mirror of
https://github.com/Doctorado-ML/STree.git
synced 2025-08-17 16:36:01 +00:00
Compare commits
41 Commits
enhance-pa
...
v1.2.3
| Author | SHA1 | Date | |
|---|---|---|---|
cc931d8547
|
|||
|
b044a057df
|
|||
|
fc48bc8ba4
|
|||
|
8251f07674 | ||
|
|
0b15a5af11 | ||
|
|
28d905368b | ||
|
e5d49132ec
|
|||
|
8daecc4726
|
|||
|
|
bf678df159 | ||
|
|
36b08b1bcf | ||
|
36ff3da26d
|
|||
|
|
6b281ebcc8 | ||
|
|
3aaddd096f | ||
|
|
15a5a4c407 | ||
|
|
0afe14a447 | ||
|
|
fc9b7b5c92 | ||
|
|
3f79d2877f | ||
|
ecc2800705
|
|||
|
0524d47d64
|
|||
|
d46f544466
|
|||
|
79190ef2e1
|
|||
|
|
4f04e72670 | ||
|
5cef0f4875
|
|||
|
28c7558f01
|
|||
|
|
e19d10f6a7 | ||
|
|
02de394c96 | ||
|
|
a4aac9d310 | ||
|
|
8a18c998df | ||
|
b55f59a3ec
|
|||
|
783d105099
|
|||
|
c36f685263
|
|||
|
0f89b044f1
|
|||
|
|
6ba973dfe1 | ||
|
|
460c63a6d0 | ||
|
|
f438124057 | ||
|
|
147dad684c | ||
|
|
3bdac9bd60 | ||
|
|
e4ac5075e5 | ||
|
|
36816074ff | ||
|
475ad7e752
|
|||
|
|
1c869e154e |
56
.github/workflows/codeql-analysis.yml
vendored
Normal file
56
.github/workflows/codeql-analysis.yml
vendored
Normal file
@@ -0,0 +1,56 @@
|
||||
name: "CodeQL"
|
||||
|
||||
on:
|
||||
push:
|
||||
branches: [ master ]
|
||||
pull_request:
|
||||
# The branches below must be a subset of the branches above
|
||||
branches: [ master ]
|
||||
schedule:
|
||||
- cron: '16 17 * * 3'
|
||||
|
||||
jobs:
|
||||
analyze:
|
||||
name: Analyze
|
||||
runs-on: ubuntu-latest
|
||||
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
language: [ 'python' ]
|
||||
# CodeQL supports [ 'cpp', 'csharp', 'go', 'java', 'javascript', 'python' ]
|
||||
# Learn more:
|
||||
# https://docs.github.com/en/free-pro-team@latest/github/finding-security-vulnerabilities-and-errors-in-your-code/configuring-code-scanning#changing-the-languages-that-are-analyzed
|
||||
|
||||
steps:
|
||||
- name: Checkout repository
|
||||
uses: actions/checkout@v2
|
||||
|
||||
# Initializes the CodeQL tools for scanning.
|
||||
- name: Initialize CodeQL
|
||||
uses: github/codeql-action/init@v1
|
||||
with:
|
||||
languages: ${{ matrix.language }}
|
||||
# If you wish to specify custom queries, you can do so here or in a config file.
|
||||
# By default, queries listed here will override any specified in a config file.
|
||||
# Prefix the list here with "+" to use these queries and those in the config file.
|
||||
# queries: ./path/to/local/query, your-org/your-repo/queries@main
|
||||
|
||||
# Autobuild attempts to build any compiled languages (C/C++, C#, or Java).
|
||||
# If this step fails, then you should remove it and run the build manually (see below)
|
||||
- name: Autobuild
|
||||
uses: github/codeql-action/autobuild@v1
|
||||
|
||||
# ℹ️ Command-line programs to run using the OS shell.
|
||||
# 📚 https://git.io/JvXDl
|
||||
|
||||
# ✏️ If the Autobuild fails above, remove it and uncomment the following three lines
|
||||
# and modify them (or add more) to build your code if your project
|
||||
# uses a compiled language
|
||||
|
||||
#- run: |
|
||||
# make bootstrap
|
||||
# make release
|
||||
|
||||
- name: Perform CodeQL Analysis
|
||||
uses: github/codeql-action/analyze@v1
|
||||
47
.github/workflows/main.yml
vendored
Normal file
47
.github/workflows/main.yml
vendored
Normal file
@@ -0,0 +1,47 @@
|
||||
name: CI
|
||||
|
||||
on:
|
||||
push:
|
||||
branches: [master]
|
||||
pull_request:
|
||||
branches: [master]
|
||||
workflow_dispatch:
|
||||
|
||||
jobs:
|
||||
build:
|
||||
runs-on: ${{ matrix.os }}
|
||||
strategy:
|
||||
matrix:
|
||||
os: [macos-latest, ubuntu-latest, windows-latest]
|
||||
python: [3.8]
|
||||
|
||||
steps:
|
||||
- uses: actions/checkout@v2
|
||||
- name: Set up Python ${{ matrix.python }}
|
||||
uses: actions/setup-python@v2
|
||||
with:
|
||||
python-version: ${{ matrix.python }}
|
||||
- name: Install dependencies
|
||||
run: |
|
||||
pip install -q --upgrade pip
|
||||
pip install -q -r requirements.txt
|
||||
pip install -q --upgrade codecov coverage black flake8 codacy-coverage
|
||||
- name: Lint
|
||||
run: |
|
||||
black --check --diff stree
|
||||
flake8 --count stree
|
||||
- name: Tests
|
||||
run: |
|
||||
coverage run -m unittest -v stree.tests
|
||||
coverage xml
|
||||
- name: Upload coverage to Codecov
|
||||
uses: codecov/codecov-action@v1
|
||||
with:
|
||||
token: ${{ secrets.CODECOV_TOKEN }}
|
||||
files: ./coverage.xml
|
||||
- name: Run codacy-coverage-reporter
|
||||
if: runner.os == 'Linux'
|
||||
uses: codacy/codacy-coverage-reporter-action@master
|
||||
with:
|
||||
project-token: ${{ secrets.CODACY_PROJECT_TOKEN }}
|
||||
coverage-reports: coverage.xml
|
||||
3
.gitignore
vendored
3
.gitignore
vendored
@@ -132,4 +132,5 @@ dmypy.json
|
||||
.vscode
|
||||
.pre-commit-config.yaml
|
||||
|
||||
**.csv
|
||||
**.csv
|
||||
.virtual_documents
|
||||
37
CITATION.cff
Normal file
37
CITATION.cff
Normal file
@@ -0,0 +1,37 @@
|
||||
cff-version: 1.2.0
|
||||
message: "If you use this software, please cite it as below."
|
||||
authors:
|
||||
- family-names: "Montañana"
|
||||
given-names: "Ricardo"
|
||||
orcid: "https://orcid.org/0000-0003-3242-5452"
|
||||
- family-names: "Gámez"
|
||||
given-names: "José A."
|
||||
orcid: "https://orcid.org/0000-0003-1188-1117"
|
||||
- family-names: "Puerta"
|
||||
given-names: "José M."
|
||||
orcid: "https://orcid.org/0000-0002-9164-5191"
|
||||
title: "STree"
|
||||
version: 1.0.2
|
||||
doi: 10.5281/zenodo.5504083
|
||||
date-released: 2021-11-02
|
||||
url: "https://github.com/Doctorado-ML/STree"
|
||||
preferred-citation:
|
||||
type: article
|
||||
authors:
|
||||
- family-names: "Montañana"
|
||||
given-names: "Ricardo"
|
||||
orcid: "https://orcid.org/0000-0003-3242-5452"
|
||||
- family-names: "Gámez"
|
||||
given-names: "José A."
|
||||
orcid: "https://orcid.org/0000-0003-1188-1117"
|
||||
- family-names: "Puerta"
|
||||
given-names: "José M."
|
||||
orcid: "https://orcid.org/0000-0002-9164-5191"
|
||||
doi: "10.1007/978-3-030-85713-4_6"
|
||||
journal: "Lecture Notes in Computer Science"
|
||||
month: 9
|
||||
start: 54
|
||||
end: 64
|
||||
title: "STree: A Single Multi-class Oblique Decision Tree Based on Support Vector Machines"
|
||||
volume: 12882
|
||||
year: 2021
|
||||
2
LICENSE
2
LICENSE
@@ -1,6 +1,6 @@
|
||||
MIT License
|
||||
|
||||
Copyright (c) 2020 Doctorado-ML
|
||||
Copyright (c) 2020-2021, Ricardo Montañana Gómez
|
||||
|
||||
Permission is hereby granted, free of charge, to any person obtaining a copy
|
||||
of this software and associated documentation files (the "Software"), to deal
|
||||
|
||||
50
Makefile
Normal file
50
Makefile
Normal file
@@ -0,0 +1,50 @@
|
||||
SHELL := /bin/bash
|
||||
.DEFAULT_GOAL := help
|
||||
.PHONY: coverage deps help lint push test doc build
|
||||
|
||||
coverage: ## Run tests with coverage
|
||||
coverage erase
|
||||
coverage run -m unittest -v stree.tests
|
||||
coverage report -m
|
||||
|
||||
deps: ## Install dependencies
|
||||
pip install -r requirements.txt
|
||||
|
||||
lint: ## Lint and static-check
|
||||
black stree
|
||||
flake8 stree
|
||||
mypy stree
|
||||
|
||||
push: ## Push code with tags
|
||||
git push && git push --tags
|
||||
|
||||
test: ## Run tests
|
||||
python -m unittest -v stree.tests
|
||||
|
||||
doc: ## Update documentation
|
||||
make -C docs --makefile=Makefile html
|
||||
|
||||
build: ## Build package
|
||||
rm -fr dist/*
|
||||
rm -fr build/*
|
||||
python setup.py sdist bdist_wheel
|
||||
|
||||
doc-clean: ## Update documentation
|
||||
make -C docs --makefile=Makefile clean
|
||||
|
||||
help: ## Show help message
|
||||
@IFS=$$'\n' ; \
|
||||
help_lines=(`fgrep -h "##" $(MAKEFILE_LIST) | fgrep -v fgrep | sed -e 's/\\$$//' | sed -e 's/##/:/'`); \
|
||||
printf "%s\n\n" "Usage: make [task]"; \
|
||||
printf "%-20s %s\n" "task" "help" ; \
|
||||
printf "%-20s %s\n" "------" "----" ; \
|
||||
for help_line in $${help_lines[@]}; do \
|
||||
IFS=$$':' ; \
|
||||
help_split=($$help_line) ; \
|
||||
help_command=`echo $${help_split[0]} | sed -e 's/^ *//' -e 's/ *$$//'` ; \
|
||||
help_info=`echo $${help_split[2]} | sed -e 's/^ *//' -e 's/ *$$//'` ; \
|
||||
printf '\033[36m'; \
|
||||
printf "%-20s %s" $$help_command ; \
|
||||
printf '\033[0m'; \
|
||||
printf "%s\n" $$help_info; \
|
||||
done
|
||||
62
README.md
62
README.md
@@ -1,8 +1,12 @@
|
||||
[](https://app.codeship.com/projects/399170)
|
||||
|
||||
[](https://codecov.io/gh/doctorado-ml/stree)
|
||||
[](https://www.codacy.com/gh/Doctorado-ML/STree?utm_source=github.com&utm_medium=referral&utm_content=Doctorado-ML/STree&utm_campaign=Badge_Grade)
|
||||
[](https://www.codacy.com/gh/Doctorado-ML/STree?utm_source=github.com&utm_medium=referral&utm_content=Doctorado-ML/STree&utm_campaign=Badge_Grade)
|
||||
[](https://lgtm.com/projects/g/Doctorado-ML/STree/context:python)
|
||||
[](https://badge.fury.io/py/STree)
|
||||
|
||||
[](https://zenodo.org/badge/latestdoi/262658230)
|
||||
|
||||
# Stree
|
||||
# STree
|
||||
|
||||
Oblique Tree classifier based on SVM nodes. The nodes are built and splitted with sklearn SVC models. Stree is a sklearn estimator and can be integrated in pipelines, grid searches, etc.
|
||||
|
||||
@@ -14,30 +18,62 @@ Oblique Tree classifier based on SVM nodes. The nodes are built and splitted wit
|
||||
pip install git+https://github.com/doctorado-ml/stree
|
||||
```
|
||||
|
||||
## Documentation
|
||||
|
||||
Can be found in [stree.readthedocs.io](https://stree.readthedocs.io/en/stable/)
|
||||
|
||||
## Examples
|
||||
|
||||
### Jupyter notebooks
|
||||
|
||||
* [](https://mybinder.org/v2/gh/Doctorado-ML/STree/master?urlpath=lab/tree/notebooks/benchmark.ipynb) Benchmark
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/benchmark.ipynb) Benchmark
|
||||
|
||||
* [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/benchmark.ipynb) Benchmark
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/features.ipynb) Some features
|
||||
|
||||
* [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/features.ipynb) Test features
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/gridsearch.ipynb) Gridsearch
|
||||
|
||||
* [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/adaboost.ipynb) Adaboost
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/ensemble.ipynb) Ensembles
|
||||
|
||||
* [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/gridsearch.ipynb) Gridsearch
|
||||
## Hyperparameters
|
||||
|
||||
* [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/test_graphs.ipynb) Test Graphics
|
||||
| | **Hyperparameter** | **Type/Values** | **Default** | **Meaning** |
|
||||
| --- | ------------------- | -------------------------------------------------------------- | ----------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
|
||||
| \* | C | \<float\> | 1.0 | Regularization parameter. The strength of the regularization is inversely proportional to C. Must be strictly positive. |
|
||||
| \* | kernel | {"liblinear", "linear", "poly", "rbf", "sigmoid"} | linear | Specifies the kernel type to be used in the algorithm. It must be one of ‘liblinear’, ‘linear’, ‘poly’ or ‘rbf’. liblinear uses [liblinear](https://www.csie.ntu.edu.tw/~cjlin/liblinear/) library and the rest uses [libsvm](https://www.csie.ntu.edu.tw/~cjlin/libsvm/) library through scikit-learn library |
|
||||
| \* | max_iter | \<int\> | 1e5 | Hard limit on iterations within solver, or -1 for no limit. |
|
||||
| \* | random_state | \<int\> | None | Controls the pseudo random number generation for shuffling the data for probability estimates. Ignored when probability is False.<br>Pass an int for reproducible output across multiple function calls |
|
||||
| | max_depth | \<int\> | None | Specifies the maximum depth of the tree |
|
||||
| \* | tol | \<float\> | 1e-4 | Tolerance for stopping criterion. |
|
||||
| \* | degree | \<int\> | 3 | Degree of the polynomial kernel function (‘poly’). Ignored by all other kernels. |
|
||||
| \* | gamma | {"scale", "auto"} or \<float\> | scale | Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.<br>if gamma='scale' (default) is passed then it uses 1 / (n_features \* X.var()) as value of gamma,<br>if ‘auto’, uses 1 / n_features. |
|
||||
| | split_criteria | {"impurity", "max_samples"} | impurity | Decides (just in case of a multi class classification) which column (class) use to split the dataset in a node\*\*. max_samples is incompatible with 'ovo' multiclass_strategy |
|
||||
| | criterion | {“gini”, “entropy”} | entropy | The function to measure the quality of a split (only used if max_features != num_features). <br>Supported criteria are “gini” for the Gini impurity and “entropy” for the information gain. |
|
||||
| | min_samples_split | \<int\> | 0 | The minimum number of samples required to split an internal node. 0 (default) for any |
|
||||
| | max_features | \<int\>, \<float\> <br><br>or {“auto”, “sqrt”, “log2”} | None | The number of features to consider when looking for the split:<br>If int, then consider max_features features at each split.<br>If float, then max_features is a fraction and int(max_features \* n_features) features are considered at each split.<br>If “auto”, then max_features=sqrt(n_features).<br>If “sqrt”, then max_features=sqrt(n_features).<br>If “log2”, then max_features=log2(n_features).<br>If None, then max_features=n_features. |
|
||||
| | splitter | {"best", "random", "trandom", "mutual", "cfs", "fcbf", "iwss"} | "random" | The strategy used to choose the feature set at each node (only used if max_features < num_features). Supported strategies are: **“best”**: sklearn SelectKBest algorithm is used in every node to choose the max_features best features. **“random”**: The algorithm generates 5 candidates and choose the best (max. info. gain) of them. **“trandom”**: The algorithm generates only one random combination. **"mutual"**: Chooses the best features w.r.t. their mutual info with the label. **"cfs"**: Apply Correlation-based Feature Selection. **"fcbf"**: Apply Fast Correlation-Based Filter. **"iwss"**: IWSS based algorithm |
|
||||
| | normalize | \<bool\> | False | If standardization of features should be applied on each node with the samples that reach it |
|
||||
| \* | multiclass_strategy | {"ovo", "ovr"} | "ovo" | Strategy to use with multiclass datasets, **"ovo"**: one versus one. **"ovr"**: one versus rest |
|
||||
|
||||
### Command line
|
||||
\* Hyperparameter used by the support vector classifier of every node
|
||||
|
||||
```bash
|
||||
python main.py
|
||||
```
|
||||
\*\* **Splitting in a STree node**
|
||||
|
||||
The decision function is applied to the dataset and distances from samples to hyperplanes are computed in a matrix. This matrix has as many columns as classes the samples belongs to (if more than two, i.e. multiclass classification) or 1 column if it's a binary class dataset. In binary classification only one hyperplane is computed and therefore only one column is needed to store the distances of the samples to it. If three or more classes are present in the dataset we need as many hyperplanes as classes are there, and therefore one column per hyperplane is needed.
|
||||
|
||||
In case of multiclass classification we have to decide which column take into account to make the split, that depends on hyperparameter _split_criteria_, if "impurity" is chosen then STree computes information gain of every split candidate using each column and chooses the one that maximize the information gain, otherwise STree choses the column with more samples with a predicted class (the column with more positive numbers in it).
|
||||
|
||||
Once we have the column to take into account for the split, the algorithm splits samples with positive distances to hyperplane from the rest.
|
||||
|
||||
## Tests
|
||||
|
||||
```bash
|
||||
python -m unittest -v stree.tests
|
||||
```
|
||||
|
||||
## License
|
||||
|
||||
STree is [MIT](https://github.com/doctorado-ml/stree/blob/master/LICENSE) licensed
|
||||
|
||||
## Reference
|
||||
|
||||
R. Montañana, J. A. Gámez, J. M. Puerta, "STree: a single multi-class oblique decision tree based on support vector machines.", 2021 LNAI 12882, pg. 54-64
|
||||
|
||||
10
codecov.yml
10
codecov.yml
@@ -1,12 +1,12 @@
|
||||
overage:
|
||||
coverage:
|
||||
status:
|
||||
project:
|
||||
default:
|
||||
target: 90%
|
||||
target: 100%
|
||||
comment:
|
||||
layout: "reach, diff, flags, files"
|
||||
behavior: default
|
||||
require_changes: false
|
||||
require_changes: false
|
||||
require_base: yes
|
||||
require_head: yes
|
||||
branches: null
|
||||
require_head: yes
|
||||
branches: null
|
||||
|
||||
20
docs/Makefile
Normal file
20
docs/Makefile
Normal file
@@ -0,0 +1,20 @@
|
||||
# Minimal makefile for Sphinx documentation
|
||||
#
|
||||
|
||||
# You can set these variables from the command line, and also
|
||||
# from the environment for the first two.
|
||||
SPHINXOPTS ?=
|
||||
SPHINXBUILD ?= sphinx-build
|
||||
SOURCEDIR = source
|
||||
BUILDDIR = build
|
||||
|
||||
# Put it first so that "make" without argument is like "make help".
|
||||
help:
|
||||
@$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
|
||||
|
||||
.PHONY: help Makefile
|
||||
|
||||
# Catch-all target: route all unknown targets to Sphinx using the new
|
||||
# "make mode" option. $(O) is meant as a shortcut for $(SPHINXOPTS).
|
||||
%: Makefile
|
||||
@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
|
||||
4
docs/requirements.txt
Normal file
4
docs/requirements.txt
Normal file
@@ -0,0 +1,4 @@
|
||||
sphinx
|
||||
sphinx-rtd-theme
|
||||
myst-parser
|
||||
mufs
|
||||
9
docs/source/api/Siterator.rst
Normal file
9
docs/source/api/Siterator.rst
Normal file
@@ -0,0 +1,9 @@
|
||||
Siterator
|
||||
=========
|
||||
|
||||
.. automodule:: Splitter
|
||||
.. autoclass:: Siterator
|
||||
:members:
|
||||
:undoc-members:
|
||||
:private-members:
|
||||
:show-inheritance:
|
||||
9
docs/source/api/Snode.rst
Normal file
9
docs/source/api/Snode.rst
Normal file
@@ -0,0 +1,9 @@
|
||||
Snode
|
||||
=====
|
||||
|
||||
.. automodule:: Splitter
|
||||
.. autoclass:: Snode
|
||||
:members:
|
||||
:undoc-members:
|
||||
:private-members:
|
||||
:show-inheritance:
|
||||
9
docs/source/api/Splitter.rst
Normal file
9
docs/source/api/Splitter.rst
Normal file
@@ -0,0 +1,9 @@
|
||||
Splitter
|
||||
========
|
||||
|
||||
.. automodule:: Splitter
|
||||
.. autoclass:: Splitter
|
||||
:members:
|
||||
:undoc-members:
|
||||
:private-members:
|
||||
:show-inheritance:
|
||||
9
docs/source/api/Stree.rst
Normal file
9
docs/source/api/Stree.rst
Normal file
@@ -0,0 +1,9 @@
|
||||
Stree
|
||||
=====
|
||||
|
||||
.. automodule:: stree
|
||||
.. autoclass:: Stree
|
||||
:members:
|
||||
:undoc-members:
|
||||
:private-members:
|
||||
:show-inheritance:
|
||||
11
docs/source/api/index.rst
Normal file
11
docs/source/api/index.rst
Normal file
@@ -0,0 +1,11 @@
|
||||
API index
|
||||
=========
|
||||
|
||||
.. toctree::
|
||||
:maxdepth: 2
|
||||
:caption: Contents:
|
||||
|
||||
Stree
|
||||
Siterator
|
||||
Snode
|
||||
Splitter
|
||||
57
docs/source/conf.py
Normal file
57
docs/source/conf.py
Normal file
@@ -0,0 +1,57 @@
|
||||
# Configuration file for the Sphinx documentation builder.
|
||||
#
|
||||
# This file only contains a selection of the most common options. For a full
|
||||
# list see the documentation:
|
||||
# https://www.sphinx-doc.org/en/master/usage/configuration.html
|
||||
|
||||
# -- Path setup --------------------------------------------------------------
|
||||
|
||||
# If extensions (or modules to document with autodoc) are in another directory,
|
||||
# add these directories to sys.path here. If the directory is relative to the
|
||||
# documentation root, use os.path.abspath to make it absolute, like shown here.
|
||||
#
|
||||
import os
|
||||
import sys
|
||||
import stree
|
||||
|
||||
sys.path.insert(0, os.path.abspath("../../stree/"))
|
||||
|
||||
|
||||
# -- Project information -----------------------------------------------------
|
||||
|
||||
project = "STree"
|
||||
copyright = "2020 - 2021, Ricardo Montañana Gómez"
|
||||
author = "Ricardo Montañana Gómez"
|
||||
|
||||
# The full version, including alpha/beta/rc tags
|
||||
version = stree.__version__
|
||||
release = version
|
||||
|
||||
|
||||
# -- General configuration ---------------------------------------------------
|
||||
|
||||
# Add any Sphinx extension module names here, as strings. They can be
|
||||
# extensions coming with Sphinx (named 'sphinx.ext.*') or your custom
|
||||
# ones.
|
||||
extensions = ["myst_parser", "sphinx.ext.autodoc", "sphinx.ext.viewcode"]
|
||||
|
||||
# Add any paths that contain templates here, relative to this directory.
|
||||
templates_path = ["_templates"]
|
||||
|
||||
# List of patterns, relative to source directory, that match files and
|
||||
# directories to ignore when looking for source files.
|
||||
# This pattern also affects html_static_path and html_extra_path.
|
||||
exclude_patterns = []
|
||||
|
||||
|
||||
# -- Options for HTML output -------------------------------------------------
|
||||
|
||||
# The theme to use for HTML and HTML Help pages. See the documentation for
|
||||
# a list of builtin themes.
|
||||
#
|
||||
html_theme = "sphinx_rtd_theme"
|
||||
|
||||
# Add any paths that contain custom static files (such as style sheets) here,
|
||||
# relative to this directory. They are copied after the builtin static files,
|
||||
# so a file named "default.css" will overwrite the builtin "default.css".
|
||||
html_static_path = []
|
||||
42
docs/source/example.md
Normal file
42
docs/source/example.md
Normal file
@@ -0,0 +1,42 @@
|
||||
# Examples
|
||||
|
||||
## Notebooks
|
||||
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/benchmark.ipynb) Benchmark
|
||||
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/features.ipynb) Some features
|
||||
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/gridsearch.ipynb) Gridsearch
|
||||
|
||||
- [](https://colab.research.google.com/github/Doctorado-ML/STree/blob/master/notebooks/ensemble.ipynb) Ensembles
|
||||
|
||||
## Sample Code
|
||||
|
||||
```python
|
||||
import time
|
||||
from sklearn.model_selection import train_test_split
|
||||
from sklearn.datasets import load_iris
|
||||
from stree import Stree
|
||||
|
||||
random_state = 1
|
||||
X, y = load_iris(return_X_y=True)
|
||||
Xtrain, Xtest, ytrain, ytest = train_test_split(
|
||||
X, y, test_size=0.2, random_state=random_state
|
||||
)
|
||||
now = time.time()
|
||||
print("Predicting with max_features=sqrt(n_features)")
|
||||
clf = Stree(random_state=random_state, max_features="auto")
|
||||
clf.fit(Xtrain, ytrain)
|
||||
print(f"Took {time.time() - now:.2f} seconds to train")
|
||||
print(clf)
|
||||
print(f"Classifier's accuracy (train): {clf.score(Xtrain, ytrain):.4f}")
|
||||
print(f"Classifier's accuracy (test) : {clf.score(Xtest, ytest):.4f}")
|
||||
print("=" * 40)
|
||||
print("Predicting with max_features=n_features")
|
||||
clf = Stree(random_state=random_state)
|
||||
clf.fit(Xtrain, ytrain)
|
||||
print(f"Took {time.time() - now:.2f} seconds to train")
|
||||
print(clf)
|
||||
print(f"Classifier's accuracy (train): {clf.score(Xtrain, ytrain):.4f}")
|
||||
print(f"Classifier's accuracy (test) : {clf.score(Xtest, ytest):.4f}")
|
||||
```
|
||||
BIN
docs/source/example.png
Normal file
BIN
docs/source/example.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 3.1 MiB |
29
docs/source/hyperparameters.md
Normal file
29
docs/source/hyperparameters.md
Normal file
@@ -0,0 +1,29 @@
|
||||
# Hyperparameters
|
||||
|
||||
| | **Hyperparameter** | **Type/Values** | **Default** | **Meaning** |
|
||||
| --- | ------------------- | -------------------------------------------------------------- | ----------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
|
||||
| \* | C | \<float\> | 1.0 | Regularization parameter. The strength of the regularization is inversely proportional to C. Must be strictly positive. |
|
||||
| \* | kernel | {"liblinear", "linear", "poly", "rbf", "sigmoid"} | linear | Specifies the kernel type to be used in the algorithm. It must be one of ‘liblinear’, ‘linear’, ‘poly’ or ‘rbf’. liblinear uses [liblinear](https://www.csie.ntu.edu.tw/~cjlin/liblinear/) library and the rest uses [libsvm](https://www.csie.ntu.edu.tw/~cjlin/libsvm/) library through scikit-learn library |
|
||||
| \* | max_iter | \<int\> | 1e5 | Hard limit on iterations within solver, or -1 for no limit. |
|
||||
| \* | random_state | \<int\> | None | Controls the pseudo random number generation for shuffling the data for probability estimates. Ignored when probability is False.<br>Pass an int for reproducible output across multiple function calls |
|
||||
| | max_depth | \<int\> | None | Specifies the maximum depth of the tree |
|
||||
| \* | tol | \<float\> | 1e-4 | Tolerance for stopping criterion. |
|
||||
| \* | degree | \<int\> | 3 | Degree of the polynomial kernel function (‘poly’). Ignored by all other kernels. |
|
||||
| \* | gamma | {"scale", "auto"} or \<float\> | scale | Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.<br>if gamma='scale' (default) is passed then it uses 1 / (n_features \* X.var()) as value of gamma,<br>if ‘auto’, uses 1 / n_features. |
|
||||
| | split_criteria | {"impurity", "max_samples"} | impurity | Decides (just in case of a multi class classification) which column (class) use to split the dataset in a node\*\*. max_samples is incompatible with 'ovo' multiclass_strategy |
|
||||
| | criterion | {“gini”, “entropy”} | entropy | The function to measure the quality of a split (only used if max_features != num_features). <br>Supported criteria are “gini” for the Gini impurity and “entropy” for the information gain. |
|
||||
| | min_samples_split | \<int\> | 0 | The minimum number of samples required to split an internal node. 0 (default) for any |
|
||||
| | max_features | \<int\>, \<float\> <br><br>or {“auto”, “sqrt”, “log2”} | None | The number of features to consider when looking for the split:<br>If int, then consider max_features features at each split.<br>If float, then max_features is a fraction and int(max_features \* n_features) features are considered at each split.<br>If “auto”, then max_features=sqrt(n_features).<br>If “sqrt”, then max_features=sqrt(n_features).<br>If “log2”, then max_features=log2(n_features).<br>If None, then max_features=n_features. |
|
||||
| | splitter | {"best", "random", "trandom", "mutual", "cfs", "fcbf", "iwss"} | "random" | The strategy used to choose the feature set at each node (only used if max_features < num_features). Supported strategies are: **“best”**: sklearn SelectKBest algorithm is used in every node to choose the max_features best features. **“random”**: The algorithm generates 5 candidates and choose the best (max. info. gain) of them. **“trandom”**: The algorithm generates only one random combination. **"mutual"**: Chooses the best features w.r.t. their mutual info with the label. **"cfs"**: Apply Correlation-based Feature Selection. **"fcbf"**: Apply Fast Correlation-Based Filter. **"iwss"**: IWSS based algorithm |
|
||||
| | normalize | \<bool\> | False | If standardization of features should be applied on each node with the samples that reach it |
|
||||
| \* | multiclass_strategy | {"ovo", "ovr"} | "ovo" | Strategy to use with multiclass datasets, **"ovo"**: one versus one. **"ovr"**: one versus rest |
|
||||
|
||||
\* Hyperparameter used by the support vector classifier of every node
|
||||
|
||||
\*\* **Splitting in a STree node**
|
||||
|
||||
The decision function is applied to the dataset and distances from samples to hyperplanes are computed in a matrix. This matrix has as many columns as classes the samples belongs to (if more than two, i.e. multiclass classification) or 1 column if it's a binary class dataset. In binary classification only one hyperplane is computed and therefore only one column is needed to store the distances of the samples to it. If three or more classes are present in the dataset we need as many hyperplanes as classes are there, and therefore one column per hyperplane is needed.
|
||||
|
||||
In case of multiclass classification we have to decide which column take into account to make the split, that depends on hyperparameter _split_criteria_, if "impurity" is chosen then STree computes information gain of every split candidate using each column and chooses the one that maximize the information gain, otherwise STree choses the column with more samples with a predicted class (the column with more positive numbers in it).
|
||||
|
||||
Once we have the column to take into account for the split, the algorithm splits samples with positive distances to hyperplane from the rest.
|
||||
15
docs/source/index.rst
Normal file
15
docs/source/index.rst
Normal file
@@ -0,0 +1,15 @@
|
||||
Welcome to STree's documentation!
|
||||
=================================
|
||||
|
||||
.. toctree::
|
||||
:caption: Contents:
|
||||
:titlesonly:
|
||||
|
||||
|
||||
stree
|
||||
install
|
||||
hyperparameters
|
||||
example
|
||||
api/index
|
||||
|
||||
* :ref:`genindex`
|
||||
16
docs/source/install.rst
Normal file
16
docs/source/install.rst
Normal file
@@ -0,0 +1,16 @@
|
||||
Install
|
||||
=======
|
||||
|
||||
The main stable release
|
||||
|
||||
``pip install stree``
|
||||
|
||||
or the last development branch
|
||||
|
||||
``pip install git+https://github.com/doctorado-ml/stree``
|
||||
|
||||
Tests
|
||||
*****
|
||||
|
||||
|
||||
``python -m unittest -v stree.tests``
|
||||
17
docs/source/stree.md
Normal file
17
docs/source/stree.md
Normal file
@@ -0,0 +1,17 @@
|
||||
# STree
|
||||
|
||||
|
||||
[](https://codecov.io/gh/doctorado-ml/stree)
|
||||
[](https://www.codacy.com/gh/Doctorado-ML/STree?utm_source=github.com&utm_medium=referral&utm_content=Doctorado-ML/STree&utm_campaign=Badge_Grade)
|
||||
[](https://lgtm.com/projects/g/Doctorado-ML/STree/context:python)
|
||||
[](https://badge.fury.io/py/STree)
|
||||
|
||||
[](https://zenodo.org/badge/latestdoi/262658230)
|
||||
|
||||
Oblique Tree classifier based on SVM nodes. The nodes are built and splitted with sklearn SVC models. Stree is a sklearn estimator and can be integrated in pipelines, grid searches, etc.
|
||||
|
||||
|
||||
|
||||
## License
|
||||
|
||||
STree is [MIT](https://github.com/doctorado-ml/stree/blob/master/LICENSE) licensed
|
||||
29
main.py
29
main.py
@@ -1,29 +0,0 @@
|
||||
import time
|
||||
from sklearn.model_selection import train_test_split
|
||||
from sklearn.datasets import load_iris
|
||||
from stree import Stree
|
||||
|
||||
random_state = 1
|
||||
|
||||
X, y = load_iris(return_X_y=True)
|
||||
|
||||
Xtrain, Xtest, ytrain, ytest = train_test_split(
|
||||
X, y, test_size=0.2, random_state=random_state
|
||||
)
|
||||
|
||||
now = time.time()
|
||||
print("Predicting with max_features=sqrt(n_features)")
|
||||
clf = Stree(C=0.01, random_state=random_state, max_features="auto")
|
||||
clf.fit(Xtrain, ytrain)
|
||||
print(f"Took {time.time() - now:.2f} seconds to train")
|
||||
print(clf)
|
||||
print(f"Classifier's accuracy (train): {clf.score(Xtrain, ytrain):.4f}")
|
||||
print(f"Classifier's accuracy (test) : {clf.score(Xtest, ytest):.4f}")
|
||||
print("=" * 40)
|
||||
print("Predicting with max_features=n_features")
|
||||
clf = Stree(C=0.01, random_state=random_state)
|
||||
clf.fit(Xtrain, ytrain)
|
||||
print(f"Took {time.time() - now:.2f} seconds to train")
|
||||
print(clf)
|
||||
print(f"Classifier's accuracy (train): {clf.score(Xtrain, ytrain):.4f}")
|
||||
print(f"Classifier's accuracy (test) : {clf.score(Xtest, ytest):.4f}")
|
||||
File diff suppressed because one or more lines are too long
@@ -17,35 +17,43 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#\n",
|
||||
"# Google Colab setup\n",
|
||||
"#\n",
|
||||
"#!pip install git+https://github.com/doctorado-ml/stree"
|
||||
"#!pip install git+https://github.com/doctorado-ml/stree\n",
|
||||
"!pip install pandas"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 2,
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import time\n",
|
||||
"import os\n",
|
||||
"import random\n",
|
||||
"import warnings\n",
|
||||
"import pandas as pd\n",
|
||||
"import numpy as np\n",
|
||||
"from sklearn.ensemble import AdaBoostClassifier, BaggingClassifier\n",
|
||||
"from sklearn.model_selection import train_test_split\n",
|
||||
"from stree import Stree"
|
||||
"from sklearn.exceptions import ConvergenceWarning\n",
|
||||
"from stree import Stree\n",
|
||||
"\n",
|
||||
"warnings.filterwarnings(\"ignore\", category=ConvergenceWarning)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 3,
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import os\n",
|
||||
"if not os.path.isfile('data/creditcard.csv'):\n",
|
||||
" !wget --no-check-certificate --content-disposition http://nube.jccm.es/index.php/s/Zs7SYtZQJ3RQ2H2/download\n",
|
||||
" !tar xzf creditcard.tgz"
|
||||
@@ -53,24 +61,15 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 4,
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"tags": []
|
||||
},
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": "Fraud: 0.173% 492\nValid: 99.827% 284315\nX.shape (100492, 28) y.shape (100492,)\nFraud: 0.644% 647\nValid: 99.356% 99845\n"
|
||||
}
|
||||
],
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"random_state=1\n",
|
||||
"\n",
|
||||
"def load_creditcard(n_examples=0):\n",
|
||||
" import pandas as pd\n",
|
||||
" import numpy as np\n",
|
||||
" import random\n",
|
||||
" df = pd.read_csv('data/creditcard.csv')\n",
|
||||
" print(\"Fraud: {0:.3f}% {1}\".format(df.Class[df.Class == 1].count()*100/df.shape[0], df.Class[df.Class == 1].count()))\n",
|
||||
" print(\"Valid: {0:.3f}% {1}\".format(df.Class[df.Class == 0].count()*100/df.shape[0], df.Class[df.Class == 0].count()))\n",
|
||||
@@ -121,20 +120,14 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 5,
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"tags": []
|
||||
},
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": "Score Train: 0.9985784146480154\nScore Test: 0.9981093273185617\nTook 73.27 seconds\n"
|
||||
}
|
||||
],
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"now = time.time()\n",
|
||||
"clf = Stree(max_depth=3, random_state=random_state)\n",
|
||||
"clf = Stree(max_depth=3, random_state=random_state, max_iter=1e3)\n",
|
||||
"clf.fit(Xtrain, ytrain)\n",
|
||||
"print(\"Score Train: \", clf.score(Xtrain, ytrain))\n",
|
||||
"print(\"Score Test: \", clf.score(Xtest, ytest))\n",
|
||||
@@ -150,7 +143,7 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 6,
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
@@ -161,21 +154,15 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 7,
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"tags": []
|
||||
},
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": "Kernel: linear\tTime: 93.78 seconds\tScore Train: 0.9983083\tScore Test: 0.9983083\nKernel: rbf\tTime: 18.32 seconds\tScore Train: 0.9935602\tScore Test: 0.9935651\nKernel: poly\tTime: 69.68 seconds\tScore Train: 0.9973132\tScore Test: 0.9972801\n"
|
||||
}
|
||||
],
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"for kernel in ['linear', 'rbf', 'poly']:\n",
|
||||
" now = time.time()\n",
|
||||
" clf = AdaBoostClassifier(base_estimator=Stree(C=C, kernel=kernel, max_depth=max_depth, random_state=random_state), algorithm=\"SAMME\", n_estimators=n_estimators, random_state=random_state)\n",
|
||||
" clf = AdaBoostClassifier(base_estimator=Stree(C=C, kernel=kernel, max_depth=max_depth, random_state=random_state, max_iter=1e3), algorithm=\"SAMME\", n_estimators=n_estimators, random_state=random_state)\n",
|
||||
" clf.fit(Xtrain, ytrain)\n",
|
||||
" score_train = clf.score(Xtrain, ytrain)\n",
|
||||
" score_test = clf.score(Xtest, ytest)\n",
|
||||
@@ -191,7 +178,7 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 8,
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
@@ -202,21 +189,15 @@
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 9,
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"tags": []
|
||||
},
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": "Kernel: linear\tTime: 387.06 seconds\tScore Train: 0.9985784\tScore Test: 0.9981093\nKernel: rbf\tTime: 144.00 seconds\tScore Train: 0.9992750\tScore Test: 0.9983415\nKernel: poly\tTime: 101.78 seconds\tScore Train: 0.9992466\tScore Test: 0.9981757\n"
|
||||
}
|
||||
],
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"for kernel in ['linear', 'rbf', 'poly']:\n",
|
||||
" now = time.time()\n",
|
||||
" clf = BaggingClassifier(base_estimator=Stree(C=C, kernel=kernel, max_depth=max_depth, random_state=random_state), n_estimators=n_estimators, random_state=random_state)\n",
|
||||
" clf = BaggingClassifier(base_estimator=Stree(C=C, kernel=kernel, max_depth=max_depth, random_state=random_state, max_iter=1e3), n_estimators=n_estimators, random_state=random_state)\n",
|
||||
" clf.fit(Xtrain, ytrain)\n",
|
||||
" score_train = clf.score(Xtrain, ytrain)\n",
|
||||
" score_test = clf.score(Xtest, ytest)\n",
|
||||
@@ -225,6 +206,11 @@
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
@@ -235,14 +221,9 @@
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.7.6-final"
|
||||
},
|
||||
"orig_nbformat": 2,
|
||||
"kernelspec": {
|
||||
"name": "python37664bitgeneralvenve3128601eb614c5da59c5055670b6040",
|
||||
"display_name": "Python 3.7.6 64-bit ('general': venv)"
|
||||
"version": "3.8.2-final"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 2
|
||||
"nbformat_minor": 4
|
||||
}
|
||||
File diff suppressed because one or more lines are too long
@@ -1,247 +1,253 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Test Gridsearch\n",
|
||||
"with different kernels and different configurations"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Setup\n",
|
||||
"Uncomment the next cell if STree is not already installed"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 1,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#\n",
|
||||
"# Google Colab setup\n",
|
||||
"#\n",
|
||||
"#!pip install git+https://github.com/doctorado-ml/stree"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"id": "zIHKVxthDZEa",
|
||||
"colab_type": "code",
|
||||
"colab": {}
|
||||
},
|
||||
"source": [
|
||||
"from sklearn.ensemble import AdaBoostClassifier\n",
|
||||
"from sklearn.svm import LinearSVC\n",
|
||||
"from sklearn.model_selection import GridSearchCV, train_test_split\n",
|
||||
"from stree import Stree"
|
||||
],
|
||||
"execution_count": 2,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"id": "IEmq50QgDZEi",
|
||||
"colab_type": "code",
|
||||
"colab": {}
|
||||
},
|
||||
"source": [
|
||||
"import os\n",
|
||||
"if not os.path.isfile('data/creditcard.csv'):\n",
|
||||
" !wget --no-check-certificate --content-disposition http://nube.jccm.es/index.php/s/Zs7SYtZQJ3RQ2H2/download\n",
|
||||
" !tar xzf creditcard.tgz"
|
||||
],
|
||||
"execution_count": 3,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"id": "z9Q-YUfBDZEq",
|
||||
"colab_type": "code",
|
||||
"colab": {},
|
||||
"outputId": "afc822fb-f16a-4302-8a67-2b9e2880159b",
|
||||
"tags": []
|
||||
},
|
||||
"source": [
|
||||
"random_state=1\n",
|
||||
"\n",
|
||||
"def load_creditcard(n_examples=0):\n",
|
||||
" import pandas as pd\n",
|
||||
" import numpy as np\n",
|
||||
" import random\n",
|
||||
" df = pd.read_csv('data/creditcard.csv')\n",
|
||||
" print(\"Fraud: {0:.3f}% {1}\".format(df.Class[df.Class == 1].count()*100/df.shape[0], df.Class[df.Class == 1].count()))\n",
|
||||
" print(\"Valid: {0:.3f}% {1}\".format(df.Class[df.Class == 0].count()*100/df.shape[0], df.Class[df.Class == 0].count()))\n",
|
||||
" y = df.Class\n",
|
||||
" X = df.drop(['Class', 'Time', 'Amount'], axis=1).values\n",
|
||||
" if n_examples > 0:\n",
|
||||
" # Take first n_examples samples\n",
|
||||
" X = X[:n_examples, :]\n",
|
||||
" y = y[:n_examples, :]\n",
|
||||
" else:\n",
|
||||
" # Take all the positive samples with a number of random negatives\n",
|
||||
" if n_examples < 0:\n",
|
||||
" Xt = X[(y == 1).ravel()]\n",
|
||||
" yt = y[(y == 1).ravel()]\n",
|
||||
" indices = random.sample(range(X.shape[0]), -1 * n_examples)\n",
|
||||
" X = np.append(Xt, X[indices], axis=0)\n",
|
||||
" y = np.append(yt, y[indices], axis=0)\n",
|
||||
" print(\"X.shape\", X.shape, \" y.shape\", y.shape)\n",
|
||||
" print(\"Fraud: {0:.3f}% {1}\".format(len(y[y == 1])*100/X.shape[0], len(y[y == 1])))\n",
|
||||
" print(\"Valid: {0:.3f}% {1}\".format(len(y[y == 0]) * 100 / X.shape[0], len(y[y == 0])))\n",
|
||||
" Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, train_size=0.7, shuffle=True, random_state=random_state, stratify=y)\n",
|
||||
" return Xtrain, Xtest, ytrain, ytest\n",
|
||||
"\n",
|
||||
"data = load_creditcard(-1000) # Take all true samples + 1000 of the others\n",
|
||||
"# data = load_creditcard(5000) # Take the first 5000 samples\n",
|
||||
"# data = load_creditcard(0) # Take all the samples\n",
|
||||
"\n",
|
||||
"Xtrain = data[0]\n",
|
||||
"Xtest = data[1]\n",
|
||||
"ytrain = data[2]\n",
|
||||
"ytest = data[3]"
|
||||
],
|
||||
"execution_count": 4,
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": "Fraud: 0.173% 492\nValid: 99.827% 284315\nX.shape (1492, 28) y.shape (1492,)\nFraud: 32.976% 492\nValid: 67.024% 1000\n"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Tests"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"id": "HmX3kR4PDZEw",
|
||||
"colab_type": "code",
|
||||
"colab": {}
|
||||
},
|
||||
"source": [
|
||||
"parameters = {\n",
|
||||
" 'base_estimator': [Stree()],\n",
|
||||
" 'n_estimators': [10, 25],\n",
|
||||
" 'learning_rate': [.5, 1],\n",
|
||||
" 'base_estimator__tol': [.1, 1e-02],\n",
|
||||
" 'base_estimator__max_depth': [3, 5],\n",
|
||||
" 'base_estimator__C': [7, 55],\n",
|
||||
" 'base_estimator__kernel': ['linear', 'poly', 'rbf']\n",
|
||||
"}"
|
||||
],
|
||||
"execution_count": 5,
|
||||
"outputs": []
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": 6,
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "execute_result",
|
||||
"data": {
|
||||
"text/plain": "{'C': 1.0,\n 'criterion': 'gini',\n 'degree': 3,\n 'gamma': 'scale',\n 'kernel': 'linear',\n 'max_depth': None,\n 'max_features': None,\n 'max_iter': 1000,\n 'min_samples_split': 0,\n 'random_state': None,\n 'split_criteria': 'max_samples',\n 'splitter': 'random',\n 'tol': 0.0001}"
|
||||
},
|
||||
"metadata": {},
|
||||
"execution_count": 6
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"Stree().get_params()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"id": "CrcB8o6EDZE5",
|
||||
"colab_type": "code",
|
||||
"colab": {},
|
||||
"outputId": "7703413a-d563-4289-a13b-532f38f82762",
|
||||
"tags": []
|
||||
},
|
||||
"source": [
|
||||
"random_state=2020\n",
|
||||
"clf = AdaBoostClassifier(random_state=random_state, algorithm=\"SAMME\")\n",
|
||||
"grid = GridSearchCV(clf, parameters, verbose=10, n_jobs=-1, return_train_score=True)\n",
|
||||
"grid.fit(Xtrain, ytrain)"
|
||||
],
|
||||
"execution_count": 7,
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": "Fitting 5 folds for each of 96 candidates, totalling 480 fits\n[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n[Parallel(n_jobs=-1)]: Done 2 tasks | elapsed: 2.0s\n[Parallel(n_jobs=-1)]: Done 9 tasks | elapsed: 2.4s\n[Parallel(n_jobs=-1)]: Done 16 tasks | elapsed: 2.7s\n[Parallel(n_jobs=-1)]: Done 25 tasks | elapsed: 3.3s\n[Parallel(n_jobs=-1)]: Done 34 tasks | elapsed: 4.3s\n[Parallel(n_jobs=-1)]: Done 45 tasks | elapsed: 5.3s\n[Parallel(n_jobs=-1)]: Done 56 tasks | elapsed: 6.6s\n[Parallel(n_jobs=-1)]: Done 69 tasks | elapsed: 8.1s\n[Parallel(n_jobs=-1)]: Done 82 tasks | elapsed: 9.4s\n[Parallel(n_jobs=-1)]: Done 97 tasks | elapsed: 10.1s\n[Parallel(n_jobs=-1)]: Done 112 tasks | elapsed: 11.1s\n[Parallel(n_jobs=-1)]: Done 129 tasks | elapsed: 12.3s\n[Parallel(n_jobs=-1)]: Done 146 tasks | elapsed: 13.6s\n[Parallel(n_jobs=-1)]: Done 165 tasks | elapsed: 14.9s\n[Parallel(n_jobs=-1)]: Done 184 tasks | elapsed: 16.2s\n[Parallel(n_jobs=-1)]: Done 205 tasks | elapsed: 17.6s\n[Parallel(n_jobs=-1)]: Done 226 tasks | elapsed: 19.1s\n[Parallel(n_jobs=-1)]: Done 249 tasks | elapsed: 21.6s\n[Parallel(n_jobs=-1)]: Done 272 tasks | elapsed: 25.9s\n[Parallel(n_jobs=-1)]: Done 297 tasks | elapsed: 30.4s\n[Parallel(n_jobs=-1)]: Done 322 tasks | elapsed: 36.7s\n[Parallel(n_jobs=-1)]: Done 349 tasks | elapsed: 38.1s\n[Parallel(n_jobs=-1)]: Done 376 tasks | elapsed: 39.6s\n[Parallel(n_jobs=-1)]: Done 405 tasks | elapsed: 41.9s\n[Parallel(n_jobs=-1)]: Done 434 tasks | elapsed: 44.9s\n[Parallel(n_jobs=-1)]: Done 465 tasks | elapsed: 48.2s\n[Parallel(n_jobs=-1)]: Done 480 out of 480 | elapsed: 49.2s finished\n"
|
||||
},
|
||||
{
|
||||
"output_type": "execute_result",
|
||||
"data": {
|
||||
"text/plain": "GridSearchCV(estimator=AdaBoostClassifier(algorithm='SAMME', random_state=2020),\n n_jobs=-1,\n param_grid={'base_estimator': [Stree(C=55, max_depth=3, tol=0.01)],\n 'base_estimator__C': [7, 55],\n 'base_estimator__kernel': ['linear', 'poly', 'rbf'],\n 'base_estimator__max_depth': [3, 5],\n 'base_estimator__tol': [0.1, 0.01],\n 'learning_rate': [0.5, 1], 'n_estimators': [10, 25]},\n return_train_score=True, verbose=10)"
|
||||
},
|
||||
"metadata": {},
|
||||
"execution_count": 7
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"metadata": {
|
||||
"id": "ZjX88NoYDZE8",
|
||||
"colab_type": "code",
|
||||
"colab": {},
|
||||
"outputId": "285163c8-fa33-4915-8ae7-61c4f7844344",
|
||||
"tags": []
|
||||
},
|
||||
"source": [
|
||||
"print(\"Best estimator: \", grid.best_estimator_)\n",
|
||||
"print(\"Best hyperparameters: \", grid.best_params_)\n",
|
||||
"print(\"Best accuracy: \", grid.best_score_)"
|
||||
],
|
||||
"execution_count": 8,
|
||||
"outputs": [
|
||||
{
|
||||
"output_type": "stream",
|
||||
"name": "stdout",
|
||||
"text": "Best estimator: AdaBoostClassifier(algorithm='SAMME',\n base_estimator=Stree(C=55, max_depth=3, tol=0.01),\n learning_rate=0.5, n_estimators=25, random_state=2020)\nBest hyperparameters: {'base_estimator': Stree(C=55, max_depth=3, tol=0.01), 'base_estimator__C': 55, 'base_estimator__kernel': 'linear', 'base_estimator__max_depth': 3, 'base_estimator__tol': 0.01, 'learning_rate': 0.5, 'n_estimators': 25}\nBest accuracy: 0.9559440559440558\n"
|
||||
}
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.7.6-final"
|
||||
},
|
||||
"orig_nbformat": 2,
|
||||
"kernelspec": {
|
||||
"name": "python37664bitgeneralvenvfbd0a23e74cf4e778460f5ffc6761f39",
|
||||
"display_name": "Python 3.7.6 64-bit ('general': venv)"
|
||||
},
|
||||
"colab": {
|
||||
"name": "gridsearch.ipynb",
|
||||
"provenance": []
|
||||
}
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Test Gridsearch\n",
|
||||
"with different kernels and different configurations"
|
||||
]
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 0
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Setup\n",
|
||||
"Uncomment the next cell if STree is not already installed"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"#\n",
|
||||
"# Google Colab setup\n",
|
||||
"#\n",
|
||||
"#!pip install git+https://github.com/doctorado-ml/stree\n",
|
||||
"!pip install pandas"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"colab": {},
|
||||
"colab_type": "code",
|
||||
"id": "zIHKVxthDZEa"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import random\n",
|
||||
"import os\n",
|
||||
"import pandas as pd\n",
|
||||
"import numpy as np\n",
|
||||
"from sklearn.ensemble import AdaBoostClassifier\n",
|
||||
"from sklearn.svm import LinearSVC\n",
|
||||
"from sklearn.model_selection import GridSearchCV, train_test_split\n",
|
||||
"from stree import Stree"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"colab": {},
|
||||
"colab_type": "code",
|
||||
"id": "IEmq50QgDZEi"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"if not os.path.isfile('data/creditcard.csv'):\n",
|
||||
" !wget --no-check-certificate --content-disposition http://nube.jccm.es/index.php/s/Zs7SYtZQJ3RQ2H2/download\n",
|
||||
" !tar xzf creditcard.tgz"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"colab": {},
|
||||
"colab_type": "code",
|
||||
"id": "z9Q-YUfBDZEq",
|
||||
"outputId": "afc822fb-f16a-4302-8a67-2b9e2880159b",
|
||||
"tags": []
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"random_state=1\n",
|
||||
"\n",
|
||||
"def load_creditcard(n_examples=0):\n",
|
||||
" df = pd.read_csv('data/creditcard.csv')\n",
|
||||
" print(\"Fraud: {0:.3f}% {1}\".format(df.Class[df.Class == 1].count()*100/df.shape[0], df.Class[df.Class == 1].count()))\n",
|
||||
" print(\"Valid: {0:.3f}% {1}\".format(df.Class[df.Class == 0].count()*100/df.shape[0], df.Class[df.Class == 0].count()))\n",
|
||||
" y = df.Class\n",
|
||||
" X = df.drop(['Class', 'Time', 'Amount'], axis=1).values\n",
|
||||
" if n_examples > 0:\n",
|
||||
" # Take first n_examples samples\n",
|
||||
" X = X[:n_examples, :]\n",
|
||||
" y = y[:n_examples, :]\n",
|
||||
" else:\n",
|
||||
" # Take all the positive samples with a number of random negatives\n",
|
||||
" if n_examples < 0:\n",
|
||||
" Xt = X[(y == 1).ravel()]\n",
|
||||
" yt = y[(y == 1).ravel()]\n",
|
||||
" indices = random.sample(range(X.shape[0]), -1 * n_examples)\n",
|
||||
" X = np.append(Xt, X[indices], axis=0)\n",
|
||||
" y = np.append(yt, y[indices], axis=0)\n",
|
||||
" print(\"X.shape\", X.shape, \" y.shape\", y.shape)\n",
|
||||
" print(\"Fraud: {0:.3f}% {1}\".format(len(y[y == 1])*100/X.shape[0], len(y[y == 1])))\n",
|
||||
" print(\"Valid: {0:.3f}% {1}\".format(len(y[y == 0]) * 100 / X.shape[0], len(y[y == 0])))\n",
|
||||
" Xtrain, Xtest, ytrain, ytest = train_test_split(X, y, train_size=0.7, shuffle=True, random_state=random_state, stratify=y)\n",
|
||||
" return Xtrain, Xtest, ytrain, ytest\n",
|
||||
"\n",
|
||||
"data = load_creditcard(-1000) # Take all true samples + 1000 of the others\n",
|
||||
"# data = load_creditcard(5000) # Take the first 5000 samples\n",
|
||||
"# data = load_creditcard(0) # Take all the samples\n",
|
||||
"\n",
|
||||
"Xtrain = data[0]\n",
|
||||
"Xtest = data[1]\n",
|
||||
"ytrain = data[2]\n",
|
||||
"ytest = data[3]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Tests"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"colab": {},
|
||||
"colab_type": "code",
|
||||
"id": "HmX3kR4PDZEw"
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"parameters = [{\n",
|
||||
" 'base_estimator': [Stree(random_state=random_state)],\n",
|
||||
" 'n_estimators': [10, 25],\n",
|
||||
" 'learning_rate': [.5, 1],\n",
|
||||
" 'base_estimator__split_criteria': ['max_samples', 'impurity'],\n",
|
||||
" 'base_estimator__tol': [.1, 1e-02],\n",
|
||||
" 'base_estimator__max_depth': [3, 5, 7],\n",
|
||||
" 'base_estimator__C': [1, 7, 55],\n",
|
||||
" 'base_estimator__kernel': ['linear']\n",
|
||||
"},\n",
|
||||
"{\n",
|
||||
" 'base_estimator': [Stree(random_state=random_state)],\n",
|
||||
" 'n_estimators': [10, 25],\n",
|
||||
" 'learning_rate': [.5, 1],\n",
|
||||
" 'base_estimator__split_criteria': ['max_samples', 'impurity'],\n",
|
||||
" 'base_estimator__tol': [.1, 1e-02],\n",
|
||||
" 'base_estimator__max_depth': [3, 5, 7],\n",
|
||||
" 'base_estimator__C': [1, 7, 55],\n",
|
||||
" 'base_estimator__degree': [3, 5, 7],\n",
|
||||
" 'base_estimator__kernel': ['poly']\n",
|
||||
"},\n",
|
||||
"{\n",
|
||||
" 'base_estimator': [Stree(random_state=random_state)],\n",
|
||||
" 'n_estimators': [10, 25],\n",
|
||||
" 'learning_rate': [.5, 1],\n",
|
||||
" 'base_estimator__split_criteria': ['max_samples', 'impurity'],\n",
|
||||
" 'base_estimator__tol': [.1, 1e-02],\n",
|
||||
" 'base_estimator__max_depth': [3, 5, 7],\n",
|
||||
" 'base_estimator__C': [1, 7, 55],\n",
|
||||
" 'base_estimator__gamma': [.1, 1, 10],\n",
|
||||
" 'base_estimator__kernel': ['rbf']\n",
|
||||
"}]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"Stree().get_params()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"colab": {},
|
||||
"colab_type": "code",
|
||||
"id": "CrcB8o6EDZE5",
|
||||
"outputId": "7703413a-d563-4289-a13b-532f38f82762",
|
||||
"tags": []
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"clf = AdaBoostClassifier(random_state=random_state, algorithm=\"SAMME\")\n",
|
||||
"grid = GridSearchCV(clf, parameters, verbose=5, n_jobs=-1, return_train_score=True)\n",
|
||||
"grid.fit(Xtrain, ytrain)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {
|
||||
"colab": {},
|
||||
"colab_type": "code",
|
||||
"id": "ZjX88NoYDZE8",
|
||||
"outputId": "285163c8-fa33-4915-8ae7-61c4f7844344",
|
||||
"tags": []
|
||||
},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"print(\"Best estimator: \", grid.best_estimator_)\n",
|
||||
"print(\"Best hyperparameters: \", grid.best_params_)\n",
|
||||
"print(\"Best accuracy: \", grid.best_score_)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"Best estimator: AdaBoostClassifier(algorithm='SAMME',\n",
|
||||
" base_estimator=Stree(C=55, max_depth=7, random_state=1,\n",
|
||||
" split_criteria='max_samples', tol=0.1),\n",
|
||||
" learning_rate=0.5, n_estimators=25, random_state=1)\n",
|
||||
"Best hyperparameters: {'base_estimator': Stree(C=55, max_depth=7, random_state=1, split_criteria='max_samples', tol=0.1), 'base_estimator__C': 55, 'base_estimator__kernel': 'linear', 'base_estimator__max_depth': 7, 'base_estimator__split_criteria': 'max_samples', 'base_estimator__tol': 0.1, 'learning_rate': 0.5, 'n_estimators': 25}"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"Best accuracy: 0.9511777695988222"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"colab": {
|
||||
"name": "gridsearch.ipynb",
|
||||
"provenance": []
|
||||
},
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.8.2-final"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 4
|
||||
}
|
||||
@@ -1,4 +1,2 @@
|
||||
numpy
|
||||
scikit-learn
|
||||
pandas
|
||||
ipympl
|
||||
scikit-learn>0.24
|
||||
mufs
|
||||
1
runtime.txt
Normal file
1
runtime.txt
Normal file
@@ -0,0 +1 @@
|
||||
python-3.8
|
||||
38
setup.py
38
setup.py
@@ -1,36 +1,50 @@
|
||||
import setuptools
|
||||
|
||||
__version__ = "0.9rc5"
|
||||
__author__ = "Ricardo Montañana Gómez"
|
||||
|
||||
|
||||
def readme():
|
||||
with open("README.md") as f:
|
||||
return f.read()
|
||||
|
||||
|
||||
def get_data(field):
|
||||
item = ""
|
||||
with open("stree/__init__.py") as f:
|
||||
for line in f.readlines():
|
||||
if line.startswith(f"__{field}__"):
|
||||
delim = '"' if '"' in line else "'"
|
||||
item = line.split(delim)[1]
|
||||
break
|
||||
else:
|
||||
raise RuntimeError(f"Unable to find {field} string.")
|
||||
return item
|
||||
|
||||
|
||||
setuptools.setup(
|
||||
name="STree",
|
||||
version=__version__,
|
||||
license="MIT License",
|
||||
version=get_data("version"),
|
||||
license=get_data("license"),
|
||||
description="Oblique decision tree with svm nodes",
|
||||
long_description=readme(),
|
||||
long_description_content_type="text/markdown",
|
||||
packages=setuptools.find_packages(),
|
||||
url="https://github.com/doctorado-ml/stree",
|
||||
author=__author__,
|
||||
author_email="ricardo.montanana@alu.uclm.es",
|
||||
url="https://github.com/Doctorado-ML/STree#stree",
|
||||
project_urls={
|
||||
"Code": "https://github.com/Doctorado-ML/STree",
|
||||
"Documentation": "https://stree.readthedocs.io/en/latest/index.html",
|
||||
},
|
||||
author=get_data("author"),
|
||||
author_email=get_data("author_email"),
|
||||
keywords="scikit-learn oblique-classifier oblique-decision-tree decision-\
|
||||
tree svm svc",
|
||||
classifiers=[
|
||||
"Development Status :: 4 - Beta",
|
||||
"License :: OSI Approved :: MIT License",
|
||||
"Programming Language :: Python :: 3.7",
|
||||
"Development Status :: 5 - Production/Stable",
|
||||
"License :: OSI Approved :: " + get_data("license"),
|
||||
"Programming Language :: Python :: 3.8",
|
||||
"Natural Language :: English",
|
||||
"Topic :: Scientific/Engineering :: Artificial Intelligence",
|
||||
"Intended Audience :: Science/Research",
|
||||
],
|
||||
install_requires=["scikit-learn>=0.23.0", "numpy", "ipympl"],
|
||||
install_requires=["scikit-learn", "mufs"],
|
||||
test_suite="stree.tests",
|
||||
zip_safe=False,
|
||||
)
|
||||
|
||||
10
stree/.readthedocs.yaml
Normal file
10
stree/.readthedocs.yaml
Normal file
@@ -0,0 +1,10 @@
|
||||
version: 2
|
||||
|
||||
sphinx:
|
||||
configuration: docs/source/conf.py
|
||||
|
||||
python:
|
||||
version: 3.8
|
||||
install:
|
||||
- requirements: requirements.txt
|
||||
- requirements: docs/requirements.txt
|
||||
787
stree/Splitter.py
Normal file
787
stree/Splitter.py
Normal file
@@ -0,0 +1,787 @@
|
||||
"""
|
||||
Oblique decision tree classifier based on SVM nodes
|
||||
Splitter class
|
||||
"""
|
||||
|
||||
import os
|
||||
import warnings
|
||||
import random
|
||||
from math import log, factorial
|
||||
import numpy as np
|
||||
from sklearn.feature_selection import SelectKBest, mutual_info_classif
|
||||
from sklearn.preprocessing import StandardScaler
|
||||
from sklearn.svm import SVC
|
||||
from sklearn.exceptions import ConvergenceWarning
|
||||
from mufs import MUFS
|
||||
|
||||
|
||||
class Snode:
|
||||
"""
|
||||
Nodes of the tree that keeps the svm classifier and if testing the
|
||||
dataset assigned to it
|
||||
|
||||
Parameters
|
||||
----------
|
||||
clf : SVC
|
||||
Classifier used
|
||||
X : np.ndarray
|
||||
input dataset in train time (only in testing)
|
||||
y : np.ndarray
|
||||
input labes in train time
|
||||
features : np.array
|
||||
features used to compute hyperplane
|
||||
impurity : float
|
||||
impurity of the node
|
||||
title : str
|
||||
label describing the route to the node
|
||||
weight : np.ndarray, optional
|
||||
weights applied to input dataset in train time, by default None
|
||||
scaler : StandardScaler, optional
|
||||
scaler used if any, by default None
|
||||
"""
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
clf: SVC,
|
||||
X: np.ndarray,
|
||||
y: np.ndarray,
|
||||
features: np.array,
|
||||
impurity: float,
|
||||
title: str,
|
||||
weight: np.ndarray = None,
|
||||
scaler: StandardScaler = None,
|
||||
):
|
||||
self._clf = clf
|
||||
self._title = title
|
||||
self._belief = 0.0
|
||||
# Only store dataset in Testing
|
||||
self._X = X if os.environ.get("TESTING", "NS") != "NS" else None
|
||||
self._y = y
|
||||
self._down = None
|
||||
self._up = None
|
||||
self._class = None
|
||||
self._feature = None
|
||||
self._sample_weight = (
|
||||
weight if os.environ.get("TESTING", "NS") != "NS" else None
|
||||
)
|
||||
self._features = features
|
||||
self._impurity = impurity
|
||||
self._partition_column: int = -1
|
||||
self._scaler = scaler
|
||||
|
||||
@classmethod
|
||||
def copy(cls, node: "Snode") -> "Snode":
|
||||
return cls(
|
||||
node._clf,
|
||||
node._X,
|
||||
node._y,
|
||||
node._features,
|
||||
node._impurity,
|
||||
node._title,
|
||||
node._sample_weight,
|
||||
node._scaler,
|
||||
)
|
||||
|
||||
def set_partition_column(self, col: int):
|
||||
self._partition_column = col
|
||||
|
||||
def get_partition_column(self) -> int:
|
||||
return self._partition_column
|
||||
|
||||
def set_down(self, son):
|
||||
self._down = son
|
||||
|
||||
def set_title(self, title):
|
||||
self._title = title
|
||||
|
||||
def set_classifier(self, clf):
|
||||
self._clf = clf
|
||||
|
||||
def set_features(self, features):
|
||||
self._features = features
|
||||
|
||||
def set_impurity(self, impurity):
|
||||
self._impurity = impurity
|
||||
|
||||
def get_title(self) -> str:
|
||||
return self._title
|
||||
|
||||
def get_classifier(self) -> SVC:
|
||||
return self._clf
|
||||
|
||||
def get_impurity(self) -> float:
|
||||
return self._impurity
|
||||
|
||||
def get_features(self) -> np.array:
|
||||
return self._features
|
||||
|
||||
def set_up(self, son):
|
||||
self._up = son
|
||||
|
||||
def is_leaf(self) -> bool:
|
||||
return self._up is None and self._down is None
|
||||
|
||||
def get_down(self) -> "Snode":
|
||||
return self._down
|
||||
|
||||
def get_up(self) -> "Snode":
|
||||
return self._up
|
||||
|
||||
def make_predictor(self):
|
||||
"""Compute the class of the predictor and its belief based on the
|
||||
subdataset of the node only if it is a leaf
|
||||
"""
|
||||
if not self.is_leaf():
|
||||
return
|
||||
classes, card = np.unique(self._y, return_counts=True)
|
||||
if len(classes) > 1:
|
||||
max_card = max(card)
|
||||
self._class = classes[card == max_card][0]
|
||||
self._belief = max_card / np.sum(card)
|
||||
else:
|
||||
self._belief = 1
|
||||
try:
|
||||
self._class = classes[0]
|
||||
except IndexError:
|
||||
self._class = None
|
||||
|
||||
def __str__(self) -> str:
|
||||
count_values = np.unique(self._y, return_counts=True)
|
||||
if self.is_leaf():
|
||||
return (
|
||||
f"{self._title} - Leaf class={self._class} belief="
|
||||
f"{self._belief: .6f} impurity={self._impurity:.4f} "
|
||||
f"counts={count_values}"
|
||||
)
|
||||
return (
|
||||
f"{self._title} feaures={self._features} impurity="
|
||||
f"{self._impurity:.4f} "
|
||||
f"counts={count_values}"
|
||||
)
|
||||
|
||||
|
||||
class Siterator:
|
||||
"""Stree preorder iterator"""
|
||||
|
||||
def __init__(self, tree: Snode):
|
||||
self._stack = []
|
||||
self._push(tree)
|
||||
|
||||
def __iter__(self):
|
||||
# To complete the iterator interface
|
||||
return self
|
||||
|
||||
def _push(self, node: Snode):
|
||||
if node is not None:
|
||||
self._stack.append(node)
|
||||
|
||||
def __next__(self) -> Snode:
|
||||
if len(self._stack) == 0:
|
||||
raise StopIteration()
|
||||
node = self._stack.pop()
|
||||
self._push(node.get_up())
|
||||
self._push(node.get_down())
|
||||
return node
|
||||
|
||||
|
||||
class Splitter:
|
||||
"""
|
||||
Splits a dataset in two based on different criteria
|
||||
|
||||
Parameters
|
||||
----------
|
||||
clf : SVC, optional
|
||||
classifier, by default None
|
||||
criterion : str, optional
|
||||
The function to measure the quality of a split (only used if
|
||||
max_features != num_features). Supported criteria are “gini” for the
|
||||
Gini impurity and “entropy” for the information gain., by default
|
||||
"entropy", by default None
|
||||
feature_select : str, optional
|
||||
The strategy used to choose the feature set at each node (only used if
|
||||
max_features < num_features). Supported strategies are: “best”: sklearn
|
||||
SelectKBest algorithm is used in every node to choose the max_features
|
||||
best features. “random”: The algorithm generates 5 candidates and
|
||||
choose the best (max. info. gain) of them. “trandom”: The algorithm
|
||||
generates only one random combination. "mutual": Chooses the best
|
||||
features w.r.t. their mutual info with the label. "cfs": Apply
|
||||
Correlation-based Feature Selection. "fcbf": Apply Fast Correlation-
|
||||
Based, by default None
|
||||
criteria : str, optional
|
||||
ecides (just in case of a multi class classification) which column
|
||||
(class) use to split the dataset in a node. max_samples is
|
||||
incompatible with 'ovo' multiclass_strategy, by default None
|
||||
min_samples_split : int, optional
|
||||
The minimum number of samples required to split an internal node. 0
|
||||
(default) for any, by default None
|
||||
random_state : optional
|
||||
Controls the pseudo random number generation for shuffling the data for
|
||||
probability estimates. Ignored when probability is False.Pass an int
|
||||
for reproducible output across multiple function calls, by
|
||||
default None
|
||||
normalize : bool, optional
|
||||
If standardization of features should be applied on each node with the
|
||||
samples that reach it , by default False
|
||||
|
||||
Raises
|
||||
------
|
||||
ValueError
|
||||
clf has to be a sklearn estimator
|
||||
ValueError
|
||||
criterion must be gini or entropy
|
||||
ValueError
|
||||
criteria has to be max_samples or impurity
|
||||
ValueError
|
||||
splitter must be in {random, best, mutual, cfs, fcbf}
|
||||
"""
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
clf: SVC = None,
|
||||
criterion: str = None,
|
||||
feature_select: str = None,
|
||||
criteria: str = None,
|
||||
min_samples_split: int = None,
|
||||
random_state=None,
|
||||
normalize=False,
|
||||
):
|
||||
|
||||
self._clf = clf
|
||||
self._random_state = random_state
|
||||
if random_state is not None:
|
||||
random.seed(random_state)
|
||||
self._criterion = criterion
|
||||
self._min_samples_split = min_samples_split
|
||||
self._criteria = criteria
|
||||
self._feature_select = feature_select
|
||||
self._normalize = normalize
|
||||
|
||||
if clf is None:
|
||||
raise ValueError(f"clf has to be a sklearn estimator, got({clf})")
|
||||
|
||||
if criterion not in ["gini", "entropy"]:
|
||||
raise ValueError(
|
||||
f"criterion must be gini or entropy got({criterion})"
|
||||
)
|
||||
|
||||
if criteria not in [
|
||||
"max_samples",
|
||||
"impurity",
|
||||
]:
|
||||
raise ValueError(
|
||||
f"criteria has to be max_samples or impurity; got ({criteria})"
|
||||
)
|
||||
|
||||
if feature_select not in [
|
||||
"random",
|
||||
"trandom",
|
||||
"best",
|
||||
"mutual",
|
||||
"cfs",
|
||||
"fcbf",
|
||||
"iwss",
|
||||
]:
|
||||
raise ValueError(
|
||||
"splitter must be in {random, trandom, best, mutual, cfs, "
|
||||
"fcbf, iwss} "
|
||||
f"got ({feature_select})"
|
||||
)
|
||||
self.criterion_function = getattr(self, f"_{self._criterion}")
|
||||
self.decision_criteria = getattr(self, f"_{self._criteria}")
|
||||
self.fs_function = getattr(self, f"_fs_{self._feature_select}")
|
||||
|
||||
def _fs_random(
|
||||
self, dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Return the best of five random feature set combinations
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(< number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
# Random feature reduction
|
||||
n_features = dataset.shape[1]
|
||||
features_sets = self._generate_spaces(n_features, max_features)
|
||||
return self._select_best_set(dataset, labels, features_sets)
|
||||
|
||||
@staticmethod
|
||||
def _fs_trandom(
|
||||
dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Return the a random feature set combination
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(< number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
# Random feature reduction
|
||||
n_features = dataset.shape[1]
|
||||
return tuple(sorted(random.sample(range(n_features), max_features)))
|
||||
|
||||
@staticmethod
|
||||
def _fs_best(
|
||||
dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Return the variabes with higher f-score
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(< number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
return (
|
||||
SelectKBest(k=max_features)
|
||||
.fit(dataset, labels)
|
||||
.get_support(indices=True)
|
||||
)
|
||||
|
||||
def _fs_mutual(
|
||||
self, dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Return the best features with mutual information with labels
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(< number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
# return best features with mutual info with the label
|
||||
feature_list = mutual_info_classif(
|
||||
dataset, labels, random_state=self._random_state
|
||||
)
|
||||
return tuple(
|
||||
sorted(
|
||||
range(len(feature_list)), key=lambda sub: feature_list[sub]
|
||||
)[-max_features:]
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
def _fs_cfs(
|
||||
dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Correlattion-based feature selection with max_features limit
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(< number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
mufs = MUFS(max_features=max_features, discrete=False)
|
||||
return mufs.cfs(dataset, labels).get_results()
|
||||
|
||||
@staticmethod
|
||||
def _fs_fcbf(
|
||||
dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Fast Correlation-based Filter algorithm with max_features limit
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(< number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
mufs = MUFS(max_features=max_features, discrete=False)
|
||||
return mufs.fcbf(dataset, labels, 5e-4).get_results()
|
||||
|
||||
@staticmethod
|
||||
def _fs_iwss(
|
||||
dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Correlattion-based feature selection based on iwss with max_features
|
||||
limit
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(< number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
mufs = MUFS(max_features=max_features, discrete=False)
|
||||
return mufs.iwss(dataset, labels, 0.25).get_results()
|
||||
|
||||
def partition_impurity(self, y: np.array) -> np.array:
|
||||
return self.criterion_function(y)
|
||||
|
||||
@staticmethod
|
||||
def _gini(y: np.array) -> float:
|
||||
_, count = np.unique(y, return_counts=True)
|
||||
return 1 - np.sum(np.square(count / np.sum(count)))
|
||||
|
||||
@staticmethod
|
||||
def _entropy(y: np.array) -> float:
|
||||
"""Compute entropy of a labels set
|
||||
|
||||
Parameters
|
||||
----------
|
||||
y : np.array
|
||||
set of labels
|
||||
|
||||
Returns
|
||||
-------
|
||||
float
|
||||
entropy
|
||||
"""
|
||||
n_labels = len(y)
|
||||
if n_labels <= 1:
|
||||
return 0
|
||||
counts = np.bincount(y)
|
||||
proportions = counts / n_labels
|
||||
n_classes = np.count_nonzero(proportions)
|
||||
if n_classes <= 1:
|
||||
return 0
|
||||
entropy = 0.0
|
||||
# Compute standard entropy.
|
||||
for prop in proportions:
|
||||
if prop != 0.0:
|
||||
entropy -= prop * log(prop, n_classes)
|
||||
return entropy
|
||||
|
||||
def information_gain(
|
||||
self, labels: np.array, labels_up: np.array, labels_dn: np.array
|
||||
) -> float:
|
||||
"""Compute information gain of a split candidate
|
||||
|
||||
Parameters
|
||||
----------
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
labels_up : np.array
|
||||
labels of one side
|
||||
labels_dn : np.array
|
||||
labels on the other side
|
||||
|
||||
Returns
|
||||
-------
|
||||
float
|
||||
information gain
|
||||
"""
|
||||
imp_prev = self.criterion_function(labels)
|
||||
card_up = card_dn = imp_up = imp_dn = 0
|
||||
if labels_up is not None:
|
||||
card_up = labels_up.shape[0]
|
||||
imp_up = self.criterion_function(labels_up)
|
||||
if labels_dn is not None:
|
||||
card_dn = labels_dn.shape[0] if labels_dn is not None else 0
|
||||
imp_dn = self.criterion_function(labels_dn)
|
||||
samples = card_up + card_dn
|
||||
if samples == 0:
|
||||
return 0.0
|
||||
else:
|
||||
result = (
|
||||
imp_prev
|
||||
- (card_up / samples) * imp_up
|
||||
- (card_dn / samples) * imp_dn
|
||||
)
|
||||
return result
|
||||
|
||||
def _select_best_set(
|
||||
self, dataset: np.array, labels: np.array, features_sets: list
|
||||
) -> list:
|
||||
"""Return the best set of features among feature_sets, the criterion is
|
||||
the information gain
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples (# samples, # features)
|
||||
labels : np.array
|
||||
array of labels
|
||||
features_sets : list
|
||||
list of features sets to check
|
||||
|
||||
Returns
|
||||
-------
|
||||
list
|
||||
best feature set
|
||||
"""
|
||||
max_gain = 0
|
||||
selected = None
|
||||
warnings.filterwarnings("ignore", category=ConvergenceWarning)
|
||||
for feature_set in features_sets:
|
||||
self._clf.fit(dataset[:, feature_set], labels)
|
||||
node = Snode(
|
||||
self._clf, dataset, labels, feature_set, 0.0, "subset"
|
||||
)
|
||||
self.partition(dataset, node, train=True)
|
||||
y1, y2 = self.part(labels)
|
||||
gain = self.information_gain(labels, y1, y2)
|
||||
if gain > max_gain:
|
||||
max_gain = gain
|
||||
selected = feature_set
|
||||
return selected if selected is not None else feature_set
|
||||
|
||||
@staticmethod
|
||||
def _generate_spaces(features: int, max_features: int) -> list:
|
||||
"""Generate at most 5 feature random combinations
|
||||
|
||||
Parameters
|
||||
----------
|
||||
features : int
|
||||
number of features in each combination
|
||||
max_features : int
|
||||
number of features in dataset
|
||||
|
||||
Returns
|
||||
-------
|
||||
list
|
||||
list with up to 5 combination of features randomly selected
|
||||
"""
|
||||
comb = set()
|
||||
# Generate at most 5 combinations
|
||||
number = factorial(features) / (
|
||||
factorial(max_features) * factorial(features - max_features)
|
||||
)
|
||||
set_length = min(5, number)
|
||||
while len(comb) < set_length:
|
||||
comb.add(
|
||||
tuple(sorted(random.sample(range(features), max_features)))
|
||||
)
|
||||
return list(comb)
|
||||
|
||||
def _get_subspaces_set(
|
||||
self, dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Compute the indices of the features selected by splitter depending
|
||||
on the self._feature_select hyper parameter
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features of the subspace
|
||||
(<= number of features in dataset)
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
indices of the features selected
|
||||
"""
|
||||
# No feature reduction
|
||||
n_features = dataset.shape[1]
|
||||
if n_features == max_features:
|
||||
return tuple(range(n_features))
|
||||
# select features as selected in constructor
|
||||
return self.fs_function(dataset, labels, max_features)
|
||||
|
||||
def get_subspace(
|
||||
self, dataset: np.array, labels: np.array, max_features: int
|
||||
) -> tuple:
|
||||
"""Re3turn a subspace of the selected dataset of max_features length.
|
||||
Depending on hyperparameter
|
||||
|
||||
Parameters
|
||||
----------
|
||||
dataset : np.array
|
||||
array of samples (# samples, # features)
|
||||
labels : np.array
|
||||
labels of the dataset
|
||||
max_features : int
|
||||
number of features to form the subspace
|
||||
|
||||
Returns
|
||||
-------
|
||||
tuple
|
||||
tuple with the dataset with only the features selected and the
|
||||
indices of the features selected
|
||||
"""
|
||||
indices = self._get_subspaces_set(dataset, labels, max_features)
|
||||
return dataset[:, indices], indices
|
||||
|
||||
def _impurity(self, data: np.array, y: np.array) -> np.array:
|
||||
"""return column of dataset to be taken into account to split dataset
|
||||
|
||||
Parameters
|
||||
----------
|
||||
data : np.array
|
||||
distances to hyper plane of every class
|
||||
y : np.array
|
||||
vector of labels (classes)
|
||||
|
||||
Returns
|
||||
-------
|
||||
np.array
|
||||
column of dataset to be taken into account to split dataset
|
||||
"""
|
||||
max_gain = 0
|
||||
selected = -1
|
||||
for col in range(data.shape[1]):
|
||||
tup = y[data[:, col] > 0]
|
||||
tdn = y[data[:, col] <= 0]
|
||||
info_gain = self.information_gain(y, tup, tdn)
|
||||
if info_gain > max_gain:
|
||||
selected = col
|
||||
max_gain = info_gain
|
||||
return selected
|
||||
|
||||
@staticmethod
|
||||
def _max_samples(data: np.array, y: np.array) -> np.array:
|
||||
"""return column of dataset to be taken into account to split dataset
|
||||
|
||||
Parameters
|
||||
----------
|
||||
data : np.array
|
||||
distances to hyper plane of every class
|
||||
y : np.array
|
||||
column of dataset to be taken into account to split dataset
|
||||
|
||||
Returns
|
||||
-------
|
||||
np.array
|
||||
column of dataset to be taken into account to split dataset
|
||||
"""
|
||||
# select the class with max number of samples
|
||||
_, samples = np.unique(y, return_counts=True)
|
||||
return np.argmax(samples)
|
||||
|
||||
def partition(self, samples: np.array, node: Snode, train: bool):
|
||||
"""Set the criteria to split arrays. Compute the indices of the samples
|
||||
that should go to one side of the tree (up)
|
||||
|
||||
Parameters
|
||||
----------
|
||||
samples : np.array
|
||||
array of samples (# samples, # features)
|
||||
node : Snode
|
||||
Node of the tree where partition is going to be made
|
||||
train : bool
|
||||
Train time - True / Test time - False
|
||||
"""
|
||||
# data contains the distances of every sample to every class hyperplane
|
||||
# array of (m, nc) nc = # classes
|
||||
data = self._distances(node, samples)
|
||||
if data.shape[0] < self._min_samples_split:
|
||||
# there aren't enough samples to split
|
||||
self._up = np.ones((data.shape[0]), dtype=bool)
|
||||
return
|
||||
if data.ndim > 1:
|
||||
# split criteria for multiclass
|
||||
# Convert data to a (m, 1) array selecting values for samples
|
||||
if train:
|
||||
# in train time we have to compute the column to take into
|
||||
# account to split the dataset
|
||||
col = self.decision_criteria(data, node._y)
|
||||
node.set_partition_column(col)
|
||||
else:
|
||||
# in predcit time just use the column computed in train time
|
||||
# is taking the classifier of class <col>
|
||||
col = node.get_partition_column()
|
||||
if col == -1:
|
||||
# No partition is producing information gain
|
||||
data = np.ones(data.shape)
|
||||
data = data[:, col]
|
||||
self._up = data > 0
|
||||
|
||||
def part(self, origin: np.array) -> list:
|
||||
"""Split an array in two based on indices (self._up) and its complement
|
||||
partition has to be called first to establish up indices
|
||||
|
||||
Parameters
|
||||
----------
|
||||
origin : np.array
|
||||
dataset to split
|
||||
|
||||
Returns
|
||||
-------
|
||||
list
|
||||
list with two splits of the array
|
||||
"""
|
||||
down = ~self._up
|
||||
return [
|
||||
origin[self._up] if any(self._up) else None,
|
||||
origin[down] if any(down) else None,
|
||||
]
|
||||
|
||||
def _distances(self, node: Snode, data: np.ndarray) -> np.array:
|
||||
"""Compute distances of the samples to the hyperplane of the node
|
||||
|
||||
Parameters
|
||||
----------
|
||||
node : Snode
|
||||
node containing the svm classifier
|
||||
data : np.ndarray
|
||||
samples to compute distance to hyperplane
|
||||
|
||||
Returns
|
||||
-------
|
||||
np.array
|
||||
array of shape (m, nc) with the distances of every sample to
|
||||
the hyperplane of every class. nc = # of classes
|
||||
"""
|
||||
X_transformed = data[:, node._features]
|
||||
if self._normalize:
|
||||
X_transformed = node._scaler.transform(X_transformed)
|
||||
return node._clf.decision_function(X_transformed)
|
||||
737
stree/Strees.py
737
stree/Strees.py
@@ -1,394 +1,158 @@
|
||||
"""
|
||||
__author__ = "Ricardo Montañana Gómez"
|
||||
__copyright__ = "Copyright 2020, Ricardo Montañana Gómez"
|
||||
__license__ = "MIT"
|
||||
__version__ = "0.9"
|
||||
Build an oblique tree classifier based on SVM Trees
|
||||
Oblique decision tree classifier based on SVM nodes
|
||||
"""
|
||||
|
||||
import os
|
||||
import numbers
|
||||
import random
|
||||
import warnings
|
||||
from math import log
|
||||
from itertools import combinations
|
||||
from typing import Optional
|
||||
import numpy as np
|
||||
from sklearn.base import BaseEstimator, ClassifierMixin
|
||||
from sklearn.svm import SVC, LinearSVC
|
||||
from sklearn.utils import check_consistent_length
|
||||
from sklearn.preprocessing import StandardScaler
|
||||
from sklearn.utils.multiclass import check_classification_targets
|
||||
from sklearn.exceptions import ConvergenceWarning
|
||||
from sklearn.utils.validation import (
|
||||
check_X_y,
|
||||
check_array,
|
||||
check_is_fitted,
|
||||
_check_sample_weight,
|
||||
)
|
||||
from sklearn.metrics._classification import _weighted_sum, _check_targets
|
||||
|
||||
|
||||
class Snode:
|
||||
"""Nodes of the tree that keeps the svm classifier and if testing the
|
||||
dataset assigned to it
|
||||
"""
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
clf: SVC,
|
||||
X: np.ndarray,
|
||||
y: np.ndarray,
|
||||
features: np.array,
|
||||
impurity: float,
|
||||
title: str,
|
||||
weight: np.ndarray = None,
|
||||
):
|
||||
self._clf = clf
|
||||
self._title = title
|
||||
self._belief = 0.0
|
||||
# Only store dataset in Testing
|
||||
self._X = X if os.environ.get("TESTING", "NS") != "NS" else None
|
||||
self._y = y
|
||||
self._down = None
|
||||
self._up = None
|
||||
self._class = None
|
||||
self._feature = None
|
||||
self._sample_weight = (
|
||||
weight if os.environ.get("TESTING", "NS") != "NS" else None
|
||||
)
|
||||
self._features = features
|
||||
self._impurity = impurity
|
||||
|
||||
@classmethod
|
||||
def copy(cls, node: "Snode") -> "Snode":
|
||||
return cls(
|
||||
node._clf,
|
||||
node._X,
|
||||
node._y,
|
||||
node._features,
|
||||
node._impurity,
|
||||
node._title,
|
||||
)
|
||||
|
||||
def set_down(self, son):
|
||||
self._down = son
|
||||
|
||||
def set_up(self, son):
|
||||
self._up = son
|
||||
|
||||
def is_leaf(self) -> bool:
|
||||
return self._up is None and self._down is None
|
||||
|
||||
def get_down(self) -> "Snode":
|
||||
return self._down
|
||||
|
||||
def get_up(self) -> "Snode":
|
||||
return self._up
|
||||
|
||||
def make_predictor(self):
|
||||
"""Compute the class of the predictor and its belief based on the
|
||||
subdataset of the node only if it is a leaf
|
||||
"""
|
||||
if not self.is_leaf():
|
||||
return
|
||||
classes, card = np.unique(self._y, return_counts=True)
|
||||
if len(classes) > 1:
|
||||
max_card = max(card)
|
||||
min_card = min(card)
|
||||
self._class = classes[card == max_card][0]
|
||||
self._belief = max_card / (max_card + min_card)
|
||||
else:
|
||||
self._belief = 1
|
||||
try:
|
||||
self._class = classes[0]
|
||||
except IndexError:
|
||||
self._class = None
|
||||
|
||||
def __str__(self) -> str:
|
||||
if self.is_leaf():
|
||||
count_values = np.unique(self._y, return_counts=True)
|
||||
result = (
|
||||
f"{self._title} - Leaf class={self._class} belief="
|
||||
f"{self._belief: .6f} impurity={self._impurity:.4f} "
|
||||
f"counts={count_values}"
|
||||
)
|
||||
return result
|
||||
else:
|
||||
return (
|
||||
f"{self._title} feaures={self._features} impurity="
|
||||
f"{self._impurity:.4f}"
|
||||
)
|
||||
|
||||
|
||||
class Siterator:
|
||||
"""Stree preorder iterator
|
||||
"""
|
||||
|
||||
def __init__(self, tree: Snode):
|
||||
self._stack = []
|
||||
self._push(tree)
|
||||
|
||||
def _push(self, node: Snode):
|
||||
if node is not None:
|
||||
self._stack.append(node)
|
||||
|
||||
def __next__(self) -> Snode:
|
||||
if len(self._stack) == 0:
|
||||
raise StopIteration()
|
||||
node = self._stack.pop()
|
||||
self._push(node.get_up())
|
||||
self._push(node.get_down())
|
||||
return node
|
||||
|
||||
|
||||
class Splitter:
|
||||
def __init__(
|
||||
self,
|
||||
clf: SVC = None,
|
||||
criterion: str = None,
|
||||
splitter_type: str = None,
|
||||
criteria: str = None,
|
||||
min_samples_split: int = None,
|
||||
random_state=None,
|
||||
):
|
||||
self._clf = clf
|
||||
self._random_state = random_state
|
||||
if random_state is not None:
|
||||
random.seed(random_state)
|
||||
self._criterion = criterion
|
||||
self._min_samples_split = min_samples_split
|
||||
self._criteria = criteria
|
||||
self._splitter_type = splitter_type
|
||||
|
||||
if clf is None:
|
||||
raise ValueError(f"clf has to be a sklearn estimator, got({clf})")
|
||||
|
||||
if criterion not in ["gini", "entropy"]:
|
||||
raise ValueError(
|
||||
f"criterion must be gini or entropy got({criterion})"
|
||||
)
|
||||
|
||||
if criteria not in ["min_distance", "max_samples", "max_distance"]:
|
||||
raise ValueError(
|
||||
"split_criteria has to be min_distance "
|
||||
f"max_distance or max_samples got ({criteria})"
|
||||
)
|
||||
|
||||
if splitter_type not in ["random", "best"]:
|
||||
raise ValueError(
|
||||
f"splitter must be either random or best got({splitter_type})"
|
||||
)
|
||||
self.criterion_function = getattr(self, f"_{self._criterion}")
|
||||
self.decision_criteria = getattr(self, f"_{self._criteria}")
|
||||
|
||||
def impurity(self, y: np.array) -> np.array:
|
||||
return self.criterion_function(y)
|
||||
|
||||
@staticmethod
|
||||
def _gini(y: np.array) -> float:
|
||||
_, count = np.unique(y, return_counts=True)
|
||||
return 1 - np.sum(np.square(count / np.sum(count)))
|
||||
|
||||
@staticmethod
|
||||
def _entropy(y: np.array) -> float:
|
||||
n_labels = len(y)
|
||||
if n_labels <= 1:
|
||||
return 0
|
||||
counts = np.bincount(y)
|
||||
proportions = counts / n_labels
|
||||
n_classes = np.count_nonzero(proportions)
|
||||
if n_classes <= 1:
|
||||
return 0
|
||||
entropy = 0.0
|
||||
# Compute standard entropy.
|
||||
for prop in proportions:
|
||||
if prop != 0.0:
|
||||
entropy -= prop * log(prop, n_classes)
|
||||
return entropy
|
||||
|
||||
def information_gain(
|
||||
self, labels: np.array, labels_up: np.array, labels_dn: np.array
|
||||
) -> float:
|
||||
imp_prev = self.criterion_function(labels)
|
||||
card_up = card_dn = imp_up = imp_dn = 0
|
||||
if labels_up is not None:
|
||||
card_up = labels_up.shape[0]
|
||||
imp_up = self.criterion_function(labels_up)
|
||||
if labels_dn is not None:
|
||||
card_dn = labels_dn.shape[0] if labels_dn is not None else 0
|
||||
imp_dn = self.criterion_function(labels_dn)
|
||||
samples = card_up + card_dn
|
||||
if samples == 0:
|
||||
return 0.0
|
||||
else:
|
||||
result = (
|
||||
imp_prev
|
||||
- (card_up / samples) * imp_up
|
||||
- (card_dn / samples) * imp_dn
|
||||
)
|
||||
return result
|
||||
|
||||
def _select_best_set(
|
||||
self, dataset: np.array, labels: np.array, features_sets: list
|
||||
) -> list:
|
||||
max_gain = 0
|
||||
selected = None
|
||||
warnings.filterwarnings("ignore", category=ConvergenceWarning)
|
||||
for feature_set in features_sets:
|
||||
self._clf.fit(dataset[:, feature_set], labels)
|
||||
node = Snode(
|
||||
self._clf, dataset, labels, feature_set, 0.0, "subset"
|
||||
)
|
||||
self.partition(dataset, node)
|
||||
y1, y2 = self.part(labels)
|
||||
gain = self.information_gain(labels, y1, y2)
|
||||
if gain > max_gain:
|
||||
max_gain = gain
|
||||
selected = feature_set
|
||||
return selected if selected is not None else feature_set
|
||||
|
||||
def _get_subspaces_set(
|
||||
self, dataset: np.array, labels: np.array, max_features: int
|
||||
) -> np.array:
|
||||
features = range(dataset.shape[1])
|
||||
features_sets = list(combinations(features, max_features))
|
||||
if len(features_sets) > 1:
|
||||
if self._splitter_type == "random":
|
||||
index = random.randint(0, len(features_sets) - 1)
|
||||
return features_sets[index]
|
||||
else:
|
||||
# get only 3 sets at most
|
||||
if len(features_sets) > 3:
|
||||
features_sets = random.sample(features_sets, 3)
|
||||
return self._select_best_set(dataset, labels, features_sets)
|
||||
else:
|
||||
return features_sets[0]
|
||||
|
||||
def get_subspace(
|
||||
self, dataset: np.array, labels: np.array, max_features: int
|
||||
) -> list:
|
||||
"""Return the best subspace to make a split
|
||||
"""
|
||||
indices = self._get_subspaces_set(dataset, labels, max_features)
|
||||
return dataset[:, indices], indices
|
||||
|
||||
@staticmethod
|
||||
def _min_distance(data: np.array, _) -> np.array:
|
||||
"""Assign class to min distances
|
||||
|
||||
return a vector of classes so partition can separate class 0 from
|
||||
the rest of classes, ie. class 0 goes to one splitted node and the
|
||||
rest of classes go to the other
|
||||
:param data: distances to hyper plane of every class
|
||||
:type data: np.array (m, n_classes)
|
||||
:param _: enable call compat with other measures
|
||||
:type _: None
|
||||
:return: vector with the class assigned to each sample
|
||||
:rtype: np.array shape (m,)
|
||||
"""
|
||||
return np.argmin(data, axis=1)
|
||||
|
||||
@staticmethod
|
||||
def _max_distance(data: np.array, _) -> np.array:
|
||||
"""Assign class to max distances
|
||||
|
||||
return a vector of classes so partition can separate class 0 from
|
||||
the rest of classes, ie. class 0 goes to one splitted node and the
|
||||
rest of classes go to the other
|
||||
:param data: distances to hyper plane of every class
|
||||
:type data: np.array (m, n_classes)
|
||||
:param _: enable call compat with other measures
|
||||
:type _: None
|
||||
:return: vector with the class assigned to each sample values
|
||||
(can be 0, 1, ...)
|
||||
:rtype: np.array shape (m,)
|
||||
"""
|
||||
return np.argmax(data, axis=1)
|
||||
|
||||
@staticmethod
|
||||
def _max_samples(data: np.array, y: np.array) -> np.array:
|
||||
"""return distances of the class with more samples
|
||||
|
||||
:param data: distances to hyper plane of every class
|
||||
:type data: np.array (m, n_classes)
|
||||
:param y: vector of labels (classes)
|
||||
:type y: np.array (m,)
|
||||
:return: vector with distances to hyperplane (can be positive or neg.)
|
||||
:rtype: np.array shape (m,)
|
||||
"""
|
||||
# select the class with max number of samples
|
||||
_, samples = np.unique(y, return_counts=True)
|
||||
selected = np.argmax(samples)
|
||||
return data[:, selected]
|
||||
|
||||
def partition(self, samples: np.array, node: Snode):
|
||||
"""Set the criteria to split arrays. Compute the indices of the samples
|
||||
that should go to one side of the tree (down)
|
||||
|
||||
"""
|
||||
data = self._distances(node, samples)
|
||||
if data.shape[0] < self._min_samples_split:
|
||||
self._down = np.ones((data.shape[0]), dtype=bool)
|
||||
return
|
||||
if data.ndim > 1:
|
||||
# split criteria for multiclass
|
||||
data = self.decision_criteria(data, node._y)
|
||||
self._down = data > 0
|
||||
|
||||
@staticmethod
|
||||
def _distances(node: Snode, data: np.ndarray) -> np.array:
|
||||
"""Compute distances of the samples to the hyperplane of the node
|
||||
|
||||
:param node: node containing the svm classifier
|
||||
:type node: Snode
|
||||
:param data: samples to find out distance to hyperplane
|
||||
:type data: np.ndarray
|
||||
:return: array of shape (m, 1) with the distances of every sample to
|
||||
the hyperplane of the node
|
||||
:rtype: np.array
|
||||
"""
|
||||
return node._clf.decision_function(data[:, node._features])
|
||||
|
||||
def part(self, origin: np.array) -> list:
|
||||
"""Split an array in two based on indices (down) and its complement
|
||||
|
||||
:param origin: dataset to split
|
||||
:type origin: np.array
|
||||
:param down: indices to use to split array
|
||||
:type down: np.array
|
||||
:return: list with two splits of the array
|
||||
:rtype: list
|
||||
"""
|
||||
up = ~self._down
|
||||
return [
|
||||
origin[up] if any(up) else None,
|
||||
origin[self._down] if any(self._down) else None,
|
||||
]
|
||||
from .Splitter import Splitter, Snode, Siterator
|
||||
|
||||
|
||||
class Stree(BaseEstimator, ClassifierMixin):
|
||||
"""Estimator that is based on binary trees of svm nodes
|
||||
"""
|
||||
Estimator that is based on binary trees of svm nodes
|
||||
can deal with sample_weights in predict, used in boosting sklearn methods
|
||||
inheriting from BaseEstimator implements get_params and set_params methods
|
||||
inheriting from ClassifierMixin implement the attribute _estimator_type
|
||||
with "classifier" as value
|
||||
|
||||
Parameters
|
||||
----------
|
||||
C : float, optional
|
||||
Regularization parameter. The strength of the regularization is
|
||||
inversely proportional to C. Must be strictly positive., by default 1.0
|
||||
kernel : str, optional
|
||||
Specifies the kernel type to be used in the algorithm. It must be one
|
||||
of ‘liblinear’, ‘linear’, ‘poly’ or ‘rbf’. liblinear uses
|
||||
[liblinear](https://www.csie.ntu.edu.tw/~cjlin/liblinear/) library and
|
||||
the rest uses [libsvm](https://www.csie.ntu.edu.tw/~cjlin/libsvm/)
|
||||
library through scikit-learn library, by default "linear"
|
||||
max_iter : int, optional
|
||||
Hard limit on iterations within solver, or -1 for no limit., by default
|
||||
1e5
|
||||
random_state : int, optional
|
||||
Controls the pseudo random number generation for shuffling the data for
|
||||
probability estimates. Ignored when probability is False.Pass an int
|
||||
for reproducible output across multiple function calls, by
|
||||
default None
|
||||
max_depth : int, optional
|
||||
Specifies the maximum depth of the tree, by default None
|
||||
tol : float, optional
|
||||
Tolerance for stopping, by default 1e-4
|
||||
degree : int, optional
|
||||
Degree of the polynomial kernel function (‘poly’). Ignored by all other
|
||||
kernels., by default 3
|
||||
gamma : str, optional
|
||||
Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.if gamma='scale'
|
||||
(default) is passed then it uses 1 / (n_features * X.var()) as value
|
||||
of gamma,if ‘auto’, uses 1 / n_features., by default "scale"
|
||||
split_criteria : str, optional
|
||||
Decides (just in case of a multi class classification) which column
|
||||
(class) use to split the dataset in a node. max_samples is
|
||||
incompatible with 'ovo' multiclass_strategy, by default "impurity"
|
||||
criterion : str, optional
|
||||
The function to measure the quality of a split (only used if
|
||||
max_features != num_features). Supported criteria are “gini” for the
|
||||
Gini impurity and “entropy” for the information gain., by default
|
||||
"entropy"
|
||||
min_samples_split : int, optional
|
||||
The minimum number of samples required to split an internal node. 0
|
||||
(default) for any, by default 0
|
||||
max_features : optional
|
||||
The number of features to consider when looking for the split: If int,
|
||||
then consider max_features features at each split. If float, then
|
||||
max_features is a fraction and int(max_features * n_features) features
|
||||
are considered at each split. If “auto”, then max_features=
|
||||
sqrt(n_features). If “sqrt”, then max_features=sqrt(n_features). If
|
||||
“log2”, then max_features=log2(n_features). If None, then max_features=
|
||||
n_features., by default None
|
||||
splitter : str, optional
|
||||
The strategy used to choose the feature set at each node (only used if
|
||||
max_features < num_features). Supported strategies are: “best”: sklearn
|
||||
SelectKBest algorithm is used in every node to choose the max_features
|
||||
best features. “random”: The algorithm generates 5 candidates and
|
||||
choose the best (max. info. gain) of them. “trandom”: The algorithm
|
||||
generates only one random combination. "mutual": Chooses the best
|
||||
features w.r.t. their mutual info with the label. "cfs": Apply
|
||||
Correlation-based Feature Selection. "fcbf": Apply Fast Correlation-
|
||||
Based , by default "random"
|
||||
multiclass_strategy : str, optional
|
||||
Strategy to use with multiclass datasets, "ovo": one versus one. "ovr":
|
||||
one versus rest, by default "ovo"
|
||||
normalize : bool, optional
|
||||
If standardization of features should be applied on each node with the
|
||||
samples that reach it , by default False
|
||||
|
||||
Attributes
|
||||
----------
|
||||
classes_ : ndarray of shape (n_classes,)
|
||||
The classes labels.
|
||||
|
||||
n_classes_ : int
|
||||
The number of classes
|
||||
|
||||
n_iter_ : int
|
||||
Max number of iterations in classifier
|
||||
|
||||
depth_ : int
|
||||
Max depht of the tree
|
||||
|
||||
n_features_ : int
|
||||
The number of features when ``fit`` is performed.
|
||||
|
||||
n_features_in_ : int
|
||||
Number of features seen during :term:`fit`.
|
||||
|
||||
max_features_ : int
|
||||
Number of features to use in hyperplane computation
|
||||
|
||||
tree_ : Node
|
||||
root of the tree
|
||||
|
||||
X_ : ndarray
|
||||
points to the input dataset
|
||||
|
||||
y_ : ndarray
|
||||
points to the input labels
|
||||
|
||||
References
|
||||
----------
|
||||
R. Montañana, J. A. Gámez, J. M. Puerta, "STree: a single multi-class
|
||||
oblique decision tree based on support vector machines.", 2021 LNAI 12882
|
||||
|
||||
|
||||
"""
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
C: float = 1.0,
|
||||
kernel: str = "linear",
|
||||
max_iter: int = 1000,
|
||||
max_iter: int = 1e5,
|
||||
random_state: int = None,
|
||||
max_depth: int = None,
|
||||
tol: float = 1e-4,
|
||||
degree: int = 3,
|
||||
gamma="scale",
|
||||
split_criteria: str = "max_samples",
|
||||
criterion: str = "gini",
|
||||
split_criteria: str = "impurity",
|
||||
criterion: str = "entropy",
|
||||
min_samples_split: int = 0,
|
||||
max_features=None,
|
||||
splitter: str = "random",
|
||||
multiclass_strategy: str = "ovo",
|
||||
normalize: bool = False,
|
||||
):
|
||||
|
||||
self.max_iter = max_iter
|
||||
self.C = C
|
||||
self.kernel = kernel
|
||||
@@ -402,9 +166,12 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
self.max_features = max_features
|
||||
self.criterion = criterion
|
||||
self.splitter = splitter
|
||||
self.normalize = normalize
|
||||
self.multiclass_strategy = multiclass_strategy
|
||||
|
||||
def _more_tags(self) -> dict:
|
||||
"""Required by sklearn to supply features of the classifier
|
||||
make mandatory the labels array
|
||||
|
||||
:return: the tag required
|
||||
:rtype: dict
|
||||
@@ -416,16 +183,19 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
) -> "Stree":
|
||||
"""Build the tree based on the dataset of samples and its labels
|
||||
|
||||
:param X: dataset of samples to make predictions
|
||||
:type X: np.array
|
||||
:param y: samples labels
|
||||
:type y: np.array
|
||||
:param sample_weight: weights of the samples. Rescale C per sample.
|
||||
Hi' weights force the classifier to put more emphasis on these points
|
||||
:type sample_weight: np.array optional
|
||||
:raises ValueError: if parameters C or max_depth are out of bounds
|
||||
:return: itself to be able to chain actions: fit().predict() ...
|
||||
:rtype: Stree
|
||||
Returns
|
||||
-------
|
||||
Stree
|
||||
itself to be able to chain actions: fit().predict() ...
|
||||
|
||||
Raises
|
||||
------
|
||||
ValueError
|
||||
if C < 0
|
||||
ValueError
|
||||
if max_depth < 1
|
||||
ValueError
|
||||
if all samples have 0 or negative weights
|
||||
"""
|
||||
# Check parameters are Ok.
|
||||
if self.C < 0:
|
||||
@@ -442,21 +212,44 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
f"Maximum depth has to be greater than 1... got (max_depth=\
|
||||
{self.max_depth})"
|
||||
)
|
||||
|
||||
if self.multiclass_strategy not in ["ovr", "ovo"]:
|
||||
raise ValueError(
|
||||
"mutliclass_strategy has to be either ovr or ovo"
|
||||
f" but got {self.multiclass_strategy}"
|
||||
)
|
||||
if self.multiclass_strategy == "ovo":
|
||||
if self.kernel == "liblinear":
|
||||
raise ValueError(
|
||||
"The kernel liblinear is incompatible with ovo "
|
||||
"multiclass_strategy"
|
||||
)
|
||||
if self.split_criteria == "max_samples":
|
||||
raise ValueError(
|
||||
"The multiclass_strategy 'ovo' is incompatible with "
|
||||
"split_criteria 'max_samples'"
|
||||
)
|
||||
kernels = ["liblinear", "linear", "rbf", "poly", "sigmoid"]
|
||||
if self.kernel not in kernels:
|
||||
raise ValueError(f"Kernel {self.kernel} not in {kernels}")
|
||||
check_classification_targets(y)
|
||||
X, y = check_X_y(X, y)
|
||||
sample_weight = _check_sample_weight(
|
||||
sample_weight, X, dtype=np.float64
|
||||
)
|
||||
if not any(sample_weight):
|
||||
raise ValueError(
|
||||
"Invalid input - all samples have zero or negative weights."
|
||||
)
|
||||
check_classification_targets(y)
|
||||
# Initialize computed parameters
|
||||
self.splitter_ = Splitter(
|
||||
clf=self._build_clf(),
|
||||
criterion=self.criterion,
|
||||
splitter_type=self.splitter,
|
||||
feature_select=self.splitter,
|
||||
criteria=self.split_criteria,
|
||||
random_state=self.random_state,
|
||||
min_samples_split=self.min_samples_split,
|
||||
normalize=self.normalize,
|
||||
)
|
||||
if self.random_state is not None:
|
||||
random.seed(self.random_state)
|
||||
@@ -467,98 +260,87 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
self.n_features_ = X.shape[1]
|
||||
self.n_features_in_ = X.shape[1]
|
||||
self.max_features_ = self._initialize_max_features()
|
||||
self.tree_ = self.train(X, y, sample_weight, 1, "root")
|
||||
self._build_predictor()
|
||||
self.tree_ = self._train(X, y, sample_weight, 1, "root")
|
||||
self.X_ = X
|
||||
self.y_ = y
|
||||
return self
|
||||
|
||||
def train(
|
||||
def _train(
|
||||
self,
|
||||
X: np.ndarray,
|
||||
y: np.ndarray,
|
||||
sample_weight: np.ndarray,
|
||||
depth: int,
|
||||
title: str,
|
||||
) -> Snode:
|
||||
) -> Optional[Snode]:
|
||||
"""Recursive function to split the original dataset into predictor
|
||||
nodes (leaves)
|
||||
|
||||
:param X: samples dataset
|
||||
:type X: np.ndarray
|
||||
:param y: samples labels
|
||||
:type y: np.ndarray
|
||||
:param sample_weight: weight of samples. Rescale C per sample.
|
||||
Hi weights force the classifier to put more emphasis on these points.
|
||||
:type sample_weight: np.ndarray
|
||||
:param depth: actual depth in the tree
|
||||
:type depth: int
|
||||
:param title: description of the node
|
||||
:type title: str
|
||||
:return: binary tree
|
||||
:rtype: Snode
|
||||
Parameters
|
||||
----------
|
||||
X : np.ndarray
|
||||
samples dataset
|
||||
y : np.ndarray
|
||||
samples labels
|
||||
sample_weight : np.ndarray
|
||||
weight of samples. Rescale C per sample.
|
||||
depth : int
|
||||
actual depth in the tree
|
||||
title : str
|
||||
description of the node
|
||||
|
||||
Returns
|
||||
-------
|
||||
Optional[Snode]
|
||||
binary tree
|
||||
"""
|
||||
if depth > self.__max_depth:
|
||||
return None
|
||||
# Mask samples with 0 weight
|
||||
if any(sample_weight == 0):
|
||||
indices_zero = sample_weight == 0
|
||||
X = X[~indices_zero, :]
|
||||
y = y[~indices_zero]
|
||||
sample_weight = sample_weight[~indices_zero]
|
||||
self.depth_ = max(depth, self.depth_)
|
||||
scaler = StandardScaler()
|
||||
node = Snode(None, X, y, X.shape[1], 0.0, title, sample_weight, scaler)
|
||||
if np.unique(y).shape[0] == 1:
|
||||
# only 1 class => pure dataset
|
||||
return Snode(
|
||||
clf=None,
|
||||
X=X,
|
||||
y=y,
|
||||
features=X.shape[1],
|
||||
impurity=0.0,
|
||||
title=title + ", <pure>",
|
||||
weight=sample_weight,
|
||||
)
|
||||
node.set_title(title + ", <pure>")
|
||||
node.make_predictor()
|
||||
return node
|
||||
# Train the model
|
||||
clf = self._build_clf()
|
||||
Xs, features = self.splitter_.get_subspace(X, y, self.max_features_)
|
||||
# solve WARNING: class label 0 specified in weight is not found
|
||||
# in bagging
|
||||
if any(sample_weight == 0):
|
||||
indices = sample_weight == 0
|
||||
y_next = y[~indices]
|
||||
# touch weights if removing any class
|
||||
if np.unique(y_next).shape[0] != self.n_classes_:
|
||||
sample_weight += 1e-5
|
||||
if self.normalize:
|
||||
scaler.fit(Xs)
|
||||
Xs = scaler.transform(Xs)
|
||||
clf.fit(Xs, y, sample_weight=sample_weight)
|
||||
impurity = self.splitter_.impurity(y)
|
||||
node = Snode(clf, X, y, features, impurity, title, sample_weight)
|
||||
self.depth_ = max(depth, self.depth_)
|
||||
self.splitter_.partition(X, node)
|
||||
node.set_impurity(self.splitter_.partition_impurity(y))
|
||||
node.set_classifier(clf)
|
||||
node.set_features(features)
|
||||
self.splitter_.partition(X, node, True)
|
||||
X_U, X_D = self.splitter_.part(X)
|
||||
y_u, y_d = self.splitter_.part(y)
|
||||
sw_u, sw_d = self.splitter_.part(sample_weight)
|
||||
if X_U is None or X_D is None:
|
||||
# didn't part anything
|
||||
return Snode(
|
||||
clf,
|
||||
X,
|
||||
y,
|
||||
features=X.shape[1],
|
||||
impurity=impurity,
|
||||
title=title + ", <cgaf>",
|
||||
weight=sample_weight,
|
||||
node.set_title(title + ", <cgaf>")
|
||||
node.make_predictor()
|
||||
return node
|
||||
node.set_up(
|
||||
self._train(X_U, y_u, sw_u, depth + 1, title + f" - Up({depth+1})")
|
||||
)
|
||||
node.set_down(
|
||||
self._train(
|
||||
X_D, y_d, sw_d, depth + 1, title + f" - Down({depth+1})"
|
||||
)
|
||||
node.set_up(self.train(X_U, y_u, sw_u, depth + 1, title + " - Up"))
|
||||
node.set_down(self.train(X_D, y_d, sw_d, depth + 1, title + " - Down"))
|
||||
)
|
||||
return node
|
||||
|
||||
def _build_predictor(self):
|
||||
"""Process the leaves to make them predictors
|
||||
"""
|
||||
|
||||
def run_tree(node: Snode):
|
||||
if node.is_leaf():
|
||||
node.make_predictor()
|
||||
return
|
||||
run_tree(node.get_down())
|
||||
run_tree(node.get_up())
|
||||
|
||||
run_tree(self.tree_)
|
||||
|
||||
def _build_clf(self):
|
||||
""" Build the correct classifier for the node
|
||||
"""
|
||||
"""Build the right classifier for the node"""
|
||||
return (
|
||||
LinearSVC(
|
||||
max_iter=self.max_iter,
|
||||
@@ -566,7 +348,7 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
C=self.C,
|
||||
tol=self.tol,
|
||||
)
|
||||
if self.kernel == "linear"
|
||||
if self.kernel == "liblinear"
|
||||
else SVC(
|
||||
kernel=self.kernel,
|
||||
max_iter=self.max_iter,
|
||||
@@ -574,6 +356,8 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
C=self.C,
|
||||
gamma=self.gamma,
|
||||
degree=self.degree,
|
||||
random_state=self.random_state,
|
||||
decision_function_shape=self.multiclass_strategy,
|
||||
)
|
||||
)
|
||||
|
||||
@@ -581,12 +365,17 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
def _reorder_results(y: np.array, indices: np.array) -> np.array:
|
||||
"""Reorder an array based on the array of indices passed
|
||||
|
||||
:param y: data untidy
|
||||
:type y: np.array
|
||||
:param indices: indices used to set order
|
||||
:type indices: np.array
|
||||
:return: array y ordered
|
||||
:rtype: np.array
|
||||
Parameters
|
||||
----------
|
||||
y : np.array
|
||||
data untidy
|
||||
indices : np.array
|
||||
indices used to set order
|
||||
|
||||
Returns
|
||||
-------
|
||||
np.array
|
||||
array y ordered
|
||||
"""
|
||||
# return array of same type given in y
|
||||
y_ordered = y.copy()
|
||||
@@ -598,10 +387,22 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
def predict(self, X: np.array) -> np.array:
|
||||
"""Predict labels for each sample in dataset passed
|
||||
|
||||
:param X: dataset of samples
|
||||
:type X: np.array
|
||||
:return: array of labels
|
||||
:rtype: np.array
|
||||
Parameters
|
||||
----------
|
||||
X : np.array
|
||||
dataset of samples
|
||||
|
||||
Returns
|
||||
-------
|
||||
np.array
|
||||
array of labels
|
||||
|
||||
Raises
|
||||
------
|
||||
ValueError
|
||||
if dataset with inconsistent number of features
|
||||
NotFittedError
|
||||
if model is not fitted
|
||||
"""
|
||||
|
||||
def predict_class(
|
||||
@@ -613,7 +414,7 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
# set a class for every sample in dataset
|
||||
prediction = np.full((xp.shape[0], 1), node._class)
|
||||
return prediction, indices
|
||||
self.splitter_.partition(xp, node)
|
||||
self.splitter_.partition(xp, node, train=False)
|
||||
x_u, x_d = self.splitter_.part(xp)
|
||||
i_u, i_d = self.splitter_.part(indices)
|
||||
prx_u, prin_u = predict_class(x_u, i_u, node.get_up())
|
||||
@@ -638,38 +439,30 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
)
|
||||
return self.classes_[result]
|
||||
|
||||
def score(
|
||||
self, X: np.array, y: np.array, sample_weight: np.array = None
|
||||
) -> float:
|
||||
"""Compute accuracy of the prediction
|
||||
def nodes_leaves(self) -> tuple:
|
||||
"""Compute the number of nodes and leaves in the built tree
|
||||
|
||||
:param X: dataset of samples to make predictions
|
||||
:type X: np.array
|
||||
:param y_true: samples labels
|
||||
:type y_true: np.array
|
||||
:param sample_weight: weights of the samples. Rescale C per sample.
|
||||
Hi' weights force the classifier to put more emphasis on these points
|
||||
:type sample_weight: np.array optional
|
||||
:return: accuracy of the prediction
|
||||
:rtype: float
|
||||
Returns
|
||||
-------
|
||||
[tuple]
|
||||
tuple with the number of nodes and the number of leaves
|
||||
"""
|
||||
# sklearn check
|
||||
check_is_fitted(self)
|
||||
check_classification_targets(y)
|
||||
X, y = check_X_y(X, y)
|
||||
y_pred = self.predict(X).reshape(y.shape)
|
||||
# Compute accuracy for each possible representation
|
||||
_, y_true, y_pred = _check_targets(y, y_pred)
|
||||
check_consistent_length(y_true, y_pred, sample_weight)
|
||||
score = y_true == y_pred
|
||||
return _weighted_sum(score, sample_weight, normalize=True)
|
||||
nodes = 0
|
||||
leaves = 0
|
||||
for node in self:
|
||||
nodes += 1
|
||||
if node.is_leaf():
|
||||
leaves += 1
|
||||
return nodes, leaves
|
||||
|
||||
def __iter__(self) -> Siterator:
|
||||
"""Create an iterator to be able to visit the nodes of the tree in
|
||||
preorder, can make a list with all the nodes in preorder
|
||||
|
||||
:return: an iterator, can for i in... and list(...)
|
||||
:rtype: Siterator
|
||||
Returns
|
||||
-------
|
||||
Siterator
|
||||
an iterator, can for i in... and list(...)
|
||||
"""
|
||||
try:
|
||||
tree = self.tree_
|
||||
@@ -680,8 +473,10 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
def __str__(self) -> str:
|
||||
"""String representation of the tree
|
||||
|
||||
:return: description of nodes in the tree in preorder
|
||||
:rtype: str
|
||||
Returns
|
||||
-------
|
||||
str
|
||||
description of nodes in the tree in preorder
|
||||
"""
|
||||
output = ""
|
||||
for i in self:
|
||||
@@ -705,6 +500,12 @@ class Stree(BaseEstimator, ClassifierMixin):
|
||||
elif self.max_features is None:
|
||||
max_features = self.n_features_
|
||||
elif isinstance(self.max_features, numbers.Integral):
|
||||
if self.max_features > self.n_features_:
|
||||
raise ValueError(
|
||||
"Invalid value for max_features. "
|
||||
"It can not be greater than number of features "
|
||||
f"({self.n_features_})"
|
||||
)
|
||||
max_features = self.max_features
|
||||
else: # float
|
||||
if self.max_features > 0.0:
|
||||
|
||||
@@ -1,3 +1,10 @@
|
||||
from .Strees import Stree, Snode, Siterator, Splitter
|
||||
from .Strees import Stree, Siterator
|
||||
|
||||
__all__ = ["Stree", "Snode", "Siterator", "Splitter"]
|
||||
__version__ = "1.2.3"
|
||||
|
||||
__author__ = "Ricardo Montañana Gómez"
|
||||
__copyright__ = "Copyright 2020-2021, Ricardo Montañana Gómez"
|
||||
__license__ = "MIT License"
|
||||
__author_email__ = "ricardo.montanana@alu.uclm.es"
|
||||
|
||||
__all__ = ["Stree", "Siterator"]
|
||||
|
||||
@@ -1,16 +1,19 @@
|
||||
import os
|
||||
import unittest
|
||||
|
||||
import numpy as np
|
||||
|
||||
from stree import Stree, Snode
|
||||
from stree import Stree
|
||||
from stree.Splitter import Snode
|
||||
from .utils import load_dataset
|
||||
|
||||
|
||||
class Snode_test(unittest.TestCase):
|
||||
def __init__(self, *args, **kwargs):
|
||||
self._random_state = 1
|
||||
self._clf = Stree(random_state=self._random_state)
|
||||
self._clf = Stree(
|
||||
random_state=self._random_state,
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
)
|
||||
self._clf.fit(*load_dataset(self._random_state))
|
||||
super().__init__(*args, **kwargs)
|
||||
|
||||
@@ -40,12 +43,13 @@ class Snode_test(unittest.TestCase):
|
||||
# Check Class
|
||||
class_computed = classes[card == max_card]
|
||||
self.assertEqual(class_computed, node._class)
|
||||
# Check Partition column
|
||||
self.assertEqual(node._partition_column, -1)
|
||||
|
||||
check_leave(self._clf.tree_)
|
||||
|
||||
def test_nodes_coefs(self):
|
||||
"""Check if the nodes of the tree have the right attributes filled
|
||||
"""
|
||||
"""Check if the nodes of the tree have the right attributes filled"""
|
||||
|
||||
def run_tree(node: Snode):
|
||||
if node._belief < 1:
|
||||
@@ -54,16 +58,44 @@ class Snode_test(unittest.TestCase):
|
||||
self.assertIsNotNone(node._clf.coef_)
|
||||
if node.is_leaf():
|
||||
return
|
||||
run_tree(node.get_down())
|
||||
run_tree(node.get_up())
|
||||
run_tree(node.get_down())
|
||||
|
||||
run_tree(self._clf.tree_)
|
||||
model = Stree(self._random_state)
|
||||
model.fit(*load_dataset(self._random_state, 3, 4))
|
||||
run_tree(model.tree_)
|
||||
|
||||
def test_make_predictor_on_leaf(self):
|
||||
test = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test")
|
||||
test.make_predictor()
|
||||
self.assertEqual(1, test._class)
|
||||
self.assertEqual(0.75, test._belief)
|
||||
self.assertEqual(-1, test._partition_column)
|
||||
|
||||
def test_set_title(self):
|
||||
test = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test")
|
||||
self.assertEqual("test", test.get_title())
|
||||
test.set_title("another")
|
||||
self.assertEqual("another", test.get_title())
|
||||
|
||||
def test_set_classifier(self):
|
||||
test = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test")
|
||||
clf = Stree()
|
||||
self.assertIsNone(test.get_classifier())
|
||||
test.set_classifier(clf)
|
||||
self.assertEqual(clf, test.get_classifier())
|
||||
|
||||
def test_set_impurity(self):
|
||||
test = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test")
|
||||
self.assertEqual(0.0, test.get_impurity())
|
||||
test.set_impurity(54.7)
|
||||
self.assertEqual(54.7, test.get_impurity())
|
||||
|
||||
def test_set_features(self):
|
||||
test = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [0, 1], 0.0, "test")
|
||||
self.assertListEqual([0, 1], test.get_features())
|
||||
test.set_features([1, 2])
|
||||
self.assertListEqual([1, 2], test.get_features())
|
||||
|
||||
def test_make_predictor_on_not_leaf(self):
|
||||
test = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test")
|
||||
@@ -71,11 +103,14 @@ class Snode_test(unittest.TestCase):
|
||||
test.make_predictor()
|
||||
self.assertIsNone(test._class)
|
||||
self.assertEqual(0, test._belief)
|
||||
self.assertEqual(-1, test._partition_column)
|
||||
self.assertEqual(-1, test.get_up()._partition_column)
|
||||
|
||||
def test_make_predictor_on_leaf_bogus_data(self):
|
||||
test = Snode(None, [1, 2, 3, 4], [], [], 0.0, "test")
|
||||
test.make_predictor()
|
||||
self.assertIsNone(test._class)
|
||||
self.assertEqual(-1, test._partition_column)
|
||||
|
||||
def test_copy_node(self):
|
||||
px = [1, 2, 3, 4]
|
||||
@@ -86,3 +121,6 @@ class Snode_test(unittest.TestCase):
|
||||
self.assertListEqual(computed._y, py)
|
||||
self.assertEqual("test", computed._title)
|
||||
self.assertIsInstance(computed._clf, Stree)
|
||||
self.assertEqual(test._partition_column, computed._partition_column)
|
||||
self.assertEqual(test._sample_weight, computed._sample_weight)
|
||||
self.assertEqual(test._scaler, computed._scaler)
|
||||
|
||||
@@ -5,7 +5,8 @@ import random
|
||||
import numpy as np
|
||||
from sklearn.svm import SVC
|
||||
from sklearn.datasets import load_wine, load_iris
|
||||
from stree import Splitter
|
||||
from stree.Splitter import Splitter
|
||||
from .utils import load_dataset, load_disc_dataset
|
||||
|
||||
|
||||
class Splitter_test(unittest.TestCase):
|
||||
@@ -17,15 +18,15 @@ class Splitter_test(unittest.TestCase):
|
||||
def build(
|
||||
clf=SVC,
|
||||
min_samples_split=0,
|
||||
splitter_type="random",
|
||||
feature_select="random",
|
||||
criterion="gini",
|
||||
criteria="min_distance",
|
||||
criteria="max_samples",
|
||||
random_state=None,
|
||||
):
|
||||
return Splitter(
|
||||
clf=clf(random_state=random_state, kernel="rbf"),
|
||||
min_samples_split=min_samples_split,
|
||||
splitter_type=splitter_type,
|
||||
feature_select=feature_select,
|
||||
criterion=criterion,
|
||||
criteria=criteria,
|
||||
random_state=random_state,
|
||||
@@ -39,24 +40,20 @@ class Splitter_test(unittest.TestCase):
|
||||
with self.assertRaises(ValueError):
|
||||
self.build(criterion="duck")
|
||||
with self.assertRaises(ValueError):
|
||||
self.build(splitter_type="duck")
|
||||
self.build(feature_select="duck")
|
||||
with self.assertRaises(ValueError):
|
||||
self.build(criteria="duck")
|
||||
with self.assertRaises(ValueError):
|
||||
_ = Splitter(clf=None)
|
||||
for splitter_type in ["best", "random"]:
|
||||
for feature_select in ["best", "random"]:
|
||||
for criterion in ["gini", "entropy"]:
|
||||
for criteria in [
|
||||
"min_distance",
|
||||
"max_samples",
|
||||
"max_distance",
|
||||
]:
|
||||
for criteria in ["max_samples", "impurity"]:
|
||||
tcl = self.build(
|
||||
splitter_type=splitter_type,
|
||||
feature_select=feature_select,
|
||||
criterion=criterion,
|
||||
criteria=criteria,
|
||||
)
|
||||
self.assertEqual(splitter_type, tcl._splitter_type)
|
||||
self.assertEqual(feature_select, tcl._feature_select)
|
||||
self.assertEqual(criterion, tcl._criterion)
|
||||
self.assertEqual(criteria, tcl._criteria)
|
||||
|
||||
@@ -138,78 +135,81 @@ class Splitter_test(unittest.TestCase):
|
||||
[0.7, 0.01, -0.1],
|
||||
[0.7, -0.9, 0.5],
|
||||
[0.1, 0.2, 0.3],
|
||||
[-0.1, 0.2, 0.3],
|
||||
[-0.1, 0.2, 0.3],
|
||||
]
|
||||
)
|
||||
expected = np.array([0.2, 0.01, -0.9, 0.2])
|
||||
y = [1, 2, 1, 0]
|
||||
expected = data[:, 0]
|
||||
y = [1, 2, 1, 0, 0, 0]
|
||||
computed = tcl._max_samples(data, y)
|
||||
self.assertEqual((4,), computed.shape)
|
||||
self.assertListEqual(expected.tolist(), computed.tolist())
|
||||
self.assertEqual(0, computed)
|
||||
computed_data = data[:, computed]
|
||||
self.assertEqual((6,), computed_data.shape)
|
||||
self.assertListEqual(expected.tolist(), computed_data.tolist())
|
||||
|
||||
def test_min_distance(self):
|
||||
tcl = self.build()
|
||||
def test_impurity(self):
|
||||
tcl = self.build(criteria="impurity")
|
||||
data = np.array(
|
||||
[
|
||||
[-0.1, 0.2, -0.3],
|
||||
[0.7, 0.01, -0.1],
|
||||
[0.7, -0.9, 0.5],
|
||||
[0.1, 0.2, 0.3],
|
||||
[-0.1, 0.2, 0.3],
|
||||
[-0.1, 0.2, 0.3],
|
||||
]
|
||||
)
|
||||
expected = np.array([2, 2, 1, 0])
|
||||
computed = tcl._min_distance(data, None)
|
||||
self.assertEqual((4,), computed.shape)
|
||||
self.assertListEqual(expected.tolist(), computed.tolist())
|
||||
expected = data[:, 2]
|
||||
y = np.array([1, 2, 1, 0, 0, 0])
|
||||
computed = tcl._impurity(data, y)
|
||||
self.assertEqual(2, computed)
|
||||
computed_data = data[:, computed]
|
||||
self.assertEqual((6,), computed_data.shape)
|
||||
self.assertListEqual(expected.tolist(), computed_data.tolist())
|
||||
|
||||
def test_max_distance(self):
|
||||
tcl = self.build(criteria="max_distance")
|
||||
data = np.array(
|
||||
[
|
||||
[-0.1, 0.2, -0.3],
|
||||
[0.7, 0.01, -0.1],
|
||||
[0.7, -0.9, 0.5],
|
||||
[0.1, 0.2, 0.3],
|
||||
]
|
||||
)
|
||||
expected = np.array([1, 0, 0, 2])
|
||||
computed = tcl._max_distance(data, None)
|
||||
self.assertEqual((4,), computed.shape)
|
||||
self.assertListEqual(expected.tolist(), computed.tolist())
|
||||
def test_generate_subspaces(self):
|
||||
features = 250
|
||||
for max_features in range(2, features):
|
||||
num = len(Splitter._generate_spaces(features, max_features))
|
||||
self.assertEqual(5, num)
|
||||
self.assertEqual(3, len(Splitter._generate_spaces(3, 2)))
|
||||
self.assertEqual(4, len(Splitter._generate_spaces(4, 3)))
|
||||
|
||||
def test_best_splitter_few_sets(self):
|
||||
X, y = load_iris(return_X_y=True)
|
||||
X = np.delete(X, 3, 1)
|
||||
tcl = self.build(splitter_type="best", random_state=self._random_state)
|
||||
tcl = self.build(
|
||||
feature_select="best", random_state=self._random_state
|
||||
)
|
||||
dataset, computed = tcl.get_subspace(X, y, max_features=2)
|
||||
self.assertListEqual([0, 2], list(computed))
|
||||
self.assertListEqual(X[:, computed].tolist(), dataset.tolist())
|
||||
|
||||
def test_splitter_parameter(self):
|
||||
expected_values = [
|
||||
[2, 3, 5, 7], # best entropy min_distance
|
||||
[0, 2, 4, 5], # best entropy max_samples
|
||||
[0, 2, 8, 12], # best entropy max_distance
|
||||
[1, 2, 5, 12], # best gini min_distance
|
||||
[0, 3, 4, 10], # best gini max_samples
|
||||
[1, 2, 9, 12], # best gini max_distance
|
||||
[3, 9, 11, 12], # random entropy min_distance
|
||||
[1, 5, 6, 9], # random entropy max_samples
|
||||
[1, 2, 4, 8], # random entropy max_distance
|
||||
[2, 6, 7, 12], # random gini min_distance
|
||||
[3, 9, 10, 11], # random gini max_samples
|
||||
[2, 5, 8, 12], # random gini max_distance
|
||||
[0, 6, 11, 12], # best entropy max_samples
|
||||
[0, 6, 11, 12], # best entropy impurity
|
||||
[0, 6, 11, 12], # best gini max_samples
|
||||
[0, 6, 11, 12], # best gini impurity
|
||||
[0, 3, 8, 12], # random entropy max_samples
|
||||
[0, 3, 7, 12], # random entropy impurity
|
||||
[1, 7, 9, 12], # random gini max_samples
|
||||
[1, 5, 8, 12], # random gini impurity
|
||||
[6, 9, 11, 12], # mutual entropy max_samples
|
||||
[6, 9, 11, 12], # mutual entropy impurity
|
||||
[6, 9, 11, 12], # mutual gini max_samples
|
||||
[6, 9, 11, 12], # mutual gini impurity
|
||||
]
|
||||
X, y = load_wine(return_X_y=True)
|
||||
rn = 0
|
||||
for splitter_type in ["best", "random"]:
|
||||
for feature_select in ["best", "random", "mutual"]:
|
||||
for criterion in ["entropy", "gini"]:
|
||||
for criteria in [
|
||||
"min_distance",
|
||||
"max_samples",
|
||||
"max_distance",
|
||||
"impurity",
|
||||
]:
|
||||
tcl = self.build(
|
||||
splitter_type=splitter_type,
|
||||
feature_select=feature_select,
|
||||
criterion=criterion,
|
||||
criteria=criteria,
|
||||
)
|
||||
@@ -219,11 +219,94 @@ class Splitter_test(unittest.TestCase):
|
||||
dataset, computed = tcl.get_subspace(X, y, max_features=4)
|
||||
# print(
|
||||
# "{}, # {:7s}{:8s}{:15s}".format(
|
||||
# list(computed), splitter_type, criterion,
|
||||
# criteria,
|
||||
# list(computed),
|
||||
# feature_select,
|
||||
# criterion,
|
||||
# criteria,
|
||||
# )
|
||||
# )
|
||||
self.assertListEqual(expected, list(computed))
|
||||
self.assertListEqual(expected, sorted(list(computed)))
|
||||
self.assertListEqual(
|
||||
X[:, computed].tolist(), dataset.tolist()
|
||||
)
|
||||
|
||||
def test_get_best_subspaces(self):
|
||||
results = [
|
||||
(4, [3, 4, 11, 13]),
|
||||
(7, [1, 3, 4, 5, 11, 13, 16]),
|
||||
(9, [1, 3, 4, 5, 7, 10, 11, 13, 16]),
|
||||
]
|
||||
X, y = load_dataset(n_features=20)
|
||||
for k, expected in results:
|
||||
tcl = self.build(
|
||||
feature_select="best",
|
||||
)
|
||||
Xs, computed = tcl.get_subspace(X, y, k)
|
||||
self.assertListEqual(expected, list(computed))
|
||||
self.assertListEqual(X[:, expected].tolist(), Xs.tolist())
|
||||
|
||||
def test_get_best_subspaces_discrete(self):
|
||||
results = [
|
||||
(4, [0, 3, 16, 18]),
|
||||
(7, [0, 3, 13, 14, 16, 18, 19]),
|
||||
(9, [0, 3, 7, 13, 14, 15, 16, 18, 19]),
|
||||
]
|
||||
X, y = load_disc_dataset(n_features=20)
|
||||
for k, expected in results:
|
||||
tcl = self.build(
|
||||
feature_select="best",
|
||||
)
|
||||
Xs, computed = tcl.get_subspace(X, y, k)
|
||||
self.assertListEqual(expected, list(computed))
|
||||
self.assertListEqual(X[:, expected].tolist(), Xs.tolist())
|
||||
|
||||
def test_get_cfs_subspaces(self):
|
||||
results = [
|
||||
(4, [1, 5, 9, 12]),
|
||||
(6, [1, 5, 9, 12, 4, 2]),
|
||||
(7, [1, 5, 9, 12, 4, 2, 3]),
|
||||
]
|
||||
X, y = load_dataset(n_features=20, n_informative=7)
|
||||
for k, expected in results:
|
||||
tcl = self.build(feature_select="cfs")
|
||||
Xs, computed = tcl.get_subspace(X, y, k)
|
||||
self.assertListEqual(expected, list(computed))
|
||||
self.assertListEqual(X[:, expected].tolist(), Xs.tolist())
|
||||
|
||||
def test_get_fcbf_subspaces(self):
|
||||
results = [
|
||||
(4, [1, 5, 9, 12]),
|
||||
(6, [1, 5, 9, 12, 4, 2]),
|
||||
(7, [1, 5, 9, 12, 4, 2, 16]),
|
||||
]
|
||||
for rs, expected in results:
|
||||
X, y = load_dataset(n_features=20, n_informative=7)
|
||||
tcl = self.build(feature_select="fcbf", random_state=rs)
|
||||
Xs, computed = tcl.get_subspace(X, y, rs)
|
||||
self.assertListEqual(expected, list(computed))
|
||||
self.assertListEqual(X[:, expected].tolist(), Xs.tolist())
|
||||
|
||||
def test_get_iwss_subspaces(self):
|
||||
results = [
|
||||
(4, [1, 5, 9, 12]),
|
||||
(6, [1, 5, 9, 12, 4, 15]),
|
||||
]
|
||||
for rs, expected in results:
|
||||
X, y = load_dataset(n_features=20, n_informative=7)
|
||||
tcl = self.build(feature_select="iwss", random_state=rs)
|
||||
Xs, computed = tcl.get_subspace(X, y, rs)
|
||||
self.assertListEqual(expected, list(computed))
|
||||
self.assertListEqual(X[:, expected].tolist(), Xs.tolist())
|
||||
|
||||
def test_get_trandom_subspaces(self):
|
||||
results = [
|
||||
(4, [3, 7, 9, 12]),
|
||||
(6, [0, 1, 2, 8, 15, 18]),
|
||||
(7, [1, 2, 4, 8, 10, 12, 13]),
|
||||
]
|
||||
for rs, expected in results:
|
||||
X, y = load_dataset(n_features=20, n_informative=7)
|
||||
tcl = self.build(feature_select="trandom", random_state=rs)
|
||||
Xs, computed = tcl.get_subspace(X, y, rs)
|
||||
self.assertListEqual(expected, list(computed))
|
||||
self.assertListEqual(X[:, expected].tolist(), Xs.tolist())
|
||||
|
||||
@@ -5,28 +5,46 @@ import warnings
|
||||
import numpy as np
|
||||
from sklearn.datasets import load_iris, load_wine
|
||||
from sklearn.exceptions import ConvergenceWarning
|
||||
from sklearn.svm import LinearSVC
|
||||
|
||||
from stree import Stree, Snode
|
||||
from stree import Stree
|
||||
from stree.Splitter import Snode
|
||||
from .utils import load_dataset
|
||||
|
||||
|
||||
class Stree_test(unittest.TestCase):
|
||||
def __init__(self, *args, **kwargs):
|
||||
self._random_state = 1
|
||||
self._kernels = ["linear", "rbf", "poly"]
|
||||
self._kernels = ["liblinear", "linear", "rbf", "poly", "sigmoid"]
|
||||
super().__init__(*args, **kwargs)
|
||||
|
||||
@classmethod
|
||||
def setUp(cls):
|
||||
os.environ["TESTING"] = "1"
|
||||
|
||||
def test_valid_kernels(self):
|
||||
X, y = load_dataset()
|
||||
for kernel in self._kernels:
|
||||
clf = Stree(kernel=kernel, multiclass_strategy="ovr")
|
||||
clf.fit(X, y)
|
||||
self.assertIsNotNone(clf.tree_)
|
||||
|
||||
def test_bogus_kernel(self):
|
||||
kernel = "other"
|
||||
X, y = load_dataset()
|
||||
clf = Stree(kernel=kernel)
|
||||
with self.assertRaises(ValueError):
|
||||
clf.fit(X, y)
|
||||
|
||||
def _check_tree(self, node: Snode):
|
||||
"""Check recursively that the nodes that are not leaves have the
|
||||
correct number of labels and its sons have the right number of elements
|
||||
in their dataset
|
||||
|
||||
Arguments:
|
||||
node {Snode} -- node to check
|
||||
Parameters
|
||||
----------
|
||||
node : Snode
|
||||
node to check
|
||||
"""
|
||||
if node.is_leaf():
|
||||
return
|
||||
@@ -37,37 +55,49 @@ class Stree_test(unittest.TestCase):
|
||||
# i.e. The partition algorithm didn't forget any sample
|
||||
self.assertEqual(node._y.shape[0], y_down.shape[0] + y_up.shape[0])
|
||||
unique_y, count_y = np.unique(node._y, return_counts=True)
|
||||
_, count_d = np.unique(y_down, return_counts=True)
|
||||
_, count_u = np.unique(y_up, return_counts=True)
|
||||
labels_d, count_d = np.unique(y_down, return_counts=True)
|
||||
labels_u, count_u = np.unique(y_up, return_counts=True)
|
||||
dict_d = {label: count_d[i] for i, label in enumerate(labels_d)}
|
||||
dict_u = {label: count_u[i] for i, label in enumerate(labels_u)}
|
||||
#
|
||||
for i in unique_y:
|
||||
number_down = count_d[i]
|
||||
try:
|
||||
number_up = count_u[i]
|
||||
except IndexError:
|
||||
number_up = dict_u[i]
|
||||
except KeyError:
|
||||
number_up = 0
|
||||
try:
|
||||
number_down = dict_d[i]
|
||||
except KeyError:
|
||||
number_down = 0
|
||||
self.assertEqual(count_y[i], number_down + number_up)
|
||||
# Is the partition made the same as the prediction?
|
||||
# as the node is not a leaf...
|
||||
_, count_yp = np.unique(y_prediction, return_counts=True)
|
||||
self.assertEqual(count_yp[0], y_up.shape[0])
|
||||
self.assertEqual(count_yp[1], y_down.shape[0])
|
||||
self.assertEqual(count_yp[1], y_up.shape[0])
|
||||
self.assertEqual(count_yp[0], y_down.shape[0])
|
||||
self._check_tree(node.get_down())
|
||||
self._check_tree(node.get_up())
|
||||
|
||||
def test_build_tree(self):
|
||||
"""Check if the tree is built the same way as predictions of models
|
||||
"""
|
||||
"""Check if the tree is built the same way as predictions of models"""
|
||||
warnings.filterwarnings("ignore")
|
||||
for kernel in self._kernels:
|
||||
clf = Stree(kernel=kernel, random_state=self._random_state)
|
||||
clf = Stree(
|
||||
kernel="sigmoid",
|
||||
multiclass_strategy="ovr" if kernel == "liblinear" else "ovo",
|
||||
random_state=self._random_state,
|
||||
)
|
||||
clf.fit(*load_dataset(self._random_state))
|
||||
self._check_tree(clf.tree_)
|
||||
|
||||
def test_single_prediction(self):
|
||||
X, y = load_dataset(self._random_state)
|
||||
for kernel in self._kernels:
|
||||
clf = Stree(kernel=kernel, random_state=self._random_state)
|
||||
clf = Stree(
|
||||
kernel=kernel,
|
||||
multiclass_strategy="ovr" if kernel == "liblinear" else "ovo",
|
||||
random_state=self._random_state,
|
||||
)
|
||||
yp = clf.fit(X, y).predict((X[0, :].reshape(-1, X.shape[1])))
|
||||
self.assertEqual(yp[0], y[0])
|
||||
|
||||
@@ -75,8 +105,12 @@ class Stree_test(unittest.TestCase):
|
||||
# First 27 elements the predictions are the same as the truth
|
||||
num = 27
|
||||
X, y = load_dataset(self._random_state)
|
||||
for kernel in self._kernels:
|
||||
clf = Stree(kernel=kernel, random_state=self._random_state)
|
||||
for kernel in ["liblinear", "linear", "rbf", "poly"]:
|
||||
clf = Stree(
|
||||
kernel=kernel,
|
||||
multiclass_strategy="ovr" if kernel == "liblinear" else "ovo",
|
||||
random_state=self._random_state,
|
||||
)
|
||||
yp = clf.fit(X, y).predict(X[:num, :])
|
||||
self.assertListEqual(y[:num].tolist(), yp.tolist())
|
||||
|
||||
@@ -86,7 +120,11 @@ class Stree_test(unittest.TestCase):
|
||||
"""
|
||||
X, y = load_dataset(self._random_state)
|
||||
for kernel in self._kernels:
|
||||
clf = Stree(kernel=kernel, random_state=self._random_state)
|
||||
clf = Stree(
|
||||
kernel=kernel,
|
||||
multiclass_strategy="ovr" if kernel == "liblinear" else "ovo",
|
||||
random_state=self._random_state,
|
||||
)
|
||||
clf.fit(X, y)
|
||||
# Compute prediction line by line
|
||||
yp_line = np.array([], dtype=int)
|
||||
@@ -99,26 +137,32 @@ class Stree_test(unittest.TestCase):
|
||||
self.assertListEqual(yp_line.tolist(), yp_once.tolist())
|
||||
|
||||
def test_iterator_and_str(self):
|
||||
"""Check preorder iterator
|
||||
"""
|
||||
"""Check preorder iterator"""
|
||||
expected = [
|
||||
"root feaures=(0, 1, 2) impurity=0.5000",
|
||||
"root - Down feaures=(0, 1, 2) impurity=0.0671",
|
||||
"root - Down - Down, <cgaf> - Leaf class=1 belief= 0.975989 "
|
||||
"impurity=0.0469 counts=(array([0, 1]), array([ 17, 691]))",
|
||||
"root - Down - Up feaures=(0, 1, 2) impurity=0.3967",
|
||||
"root - Down - Up - Down, <cgaf> - Leaf class=1 belief= 0.750000 "
|
||||
"impurity=0.3750 counts=(array([0, 1]), array([1, 3]))",
|
||||
"root - Down - Up - Up, <pure> - Leaf class=0 belief= 1.000000 "
|
||||
"impurity=0.0000 counts=(array([0]), array([7]))",
|
||||
"root - Up, <cgaf> - Leaf class=0 belief= 0.928297 impurity=0.1331"
|
||||
" counts=(array([0, 1]), array([725, 56]))",
|
||||
"root feaures=(0, 1, 2) impurity=1.0000 counts=(array([0, 1]), "
|
||||
"array([750, 750]))",
|
||||
"root - Down(2), <cgaf> - Leaf class=0 belief= 0.928297 impurity="
|
||||
"0.3722 counts=(array([0, 1]), array([725, 56]))",
|
||||
"root - Up(2) feaures=(0, 1, 2) impurity=0.2178 counts=(array([0, "
|
||||
"1]), array([ 25, 694]))",
|
||||
"root - Up(2) - Down(3) feaures=(0, 1, 2) impurity=0.8454 counts="
|
||||
"(array([0, 1]), array([8, 3]))",
|
||||
"root - Up(2) - Down(3) - Down(4), <pure> - Leaf class=0 belief= "
|
||||
"1.000000 impurity=0.0000 counts=(array([0]), array([7]))",
|
||||
"root - Up(2) - Down(3) - Up(4), <cgaf> - Leaf class=1 belief= "
|
||||
"0.750000 impurity=0.8113 counts=(array([0, 1]), array([1, 3]))",
|
||||
"root - Up(2) - Up(3), <cgaf> - Leaf class=1 belief= 0.975989 "
|
||||
"impurity=0.1634 counts=(array([0, 1]), array([ 17, 691]))",
|
||||
]
|
||||
computed = []
|
||||
expected_string = ""
|
||||
clf = Stree(kernel="linear", random_state=self._random_state)
|
||||
clf = Stree(
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
random_state=self._random_state,
|
||||
)
|
||||
clf.fit(*load_dataset(self._random_state))
|
||||
for node in clf:
|
||||
for node in iter(clf):
|
||||
computed.append(str(node))
|
||||
expected_string += str(node) + "\n"
|
||||
self.assertListEqual(expected, computed)
|
||||
@@ -154,7 +198,12 @@ class Stree_test(unittest.TestCase):
|
||||
def test_check_max_depth(self):
|
||||
depths = (3, 4)
|
||||
for depth in depths:
|
||||
tcl = Stree(random_state=self._random_state, max_depth=depth)
|
||||
tcl = Stree(
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
random_state=self._random_state,
|
||||
max_depth=depth,
|
||||
)
|
||||
tcl.fit(*load_dataset(self._random_state))
|
||||
self.assertEqual(depth, tcl.depth_)
|
||||
|
||||
@@ -175,7 +224,7 @@ class Stree_test(unittest.TestCase):
|
||||
for kernel in self._kernels:
|
||||
clf = Stree(
|
||||
kernel=kernel,
|
||||
split_criteria="max_samples",
|
||||
multiclass_strategy="ovr" if kernel == "liblinear" else "ovo",
|
||||
random_state=self._random_state,
|
||||
)
|
||||
px = [[1, 2], [5, 6], [9, 10]]
|
||||
@@ -186,47 +235,60 @@ class Stree_test(unittest.TestCase):
|
||||
self.assertListEqual(py, clf.classes_.tolist())
|
||||
|
||||
def test_muticlass_dataset(self):
|
||||
warnings.filterwarnings("ignore", category=ConvergenceWarning)
|
||||
warnings.filterwarnings("ignore", category=RuntimeWarning)
|
||||
datasets = {
|
||||
"Synt": load_dataset(random_state=self._random_state, n_classes=3),
|
||||
"Iris": load_iris(return_X_y=True),
|
||||
"Iris": load_wine(return_X_y=True),
|
||||
}
|
||||
outcomes = {
|
||||
"Synt": {
|
||||
"max_samples linear": 0.9533333333333334,
|
||||
"max_samples rbf": 0.836,
|
||||
"max_samples poly": 0.9473333333333334,
|
||||
"min_distance linear": 0.9533333333333334,
|
||||
"min_distance rbf": 0.836,
|
||||
"min_distance poly": 0.9473333333333334,
|
||||
"max_distance linear": 0.9533333333333334,
|
||||
"max_distance rbf": 0.836,
|
||||
"max_distance poly": 0.9473333333333334,
|
||||
"max_samples liblinear": 0.9493333333333334,
|
||||
"max_samples linear": 0.9426666666666667,
|
||||
"max_samples rbf": 0.9606666666666667,
|
||||
"max_samples poly": 0.9373333333333334,
|
||||
"max_samples sigmoid": 0.824,
|
||||
"impurity liblinear": 0.9493333333333334,
|
||||
"impurity linear": 0.9426666666666667,
|
||||
"impurity rbf": 0.9606666666666667,
|
||||
"impurity poly": 0.9373333333333334,
|
||||
"impurity sigmoid": 0.824,
|
||||
},
|
||||
"Iris": {
|
||||
"max_samples linear": 0.98,
|
||||
"max_samples rbf": 1.0,
|
||||
"max_samples poly": 1.0,
|
||||
"min_distance linear": 0.98,
|
||||
"min_distance rbf": 1.0,
|
||||
"min_distance poly": 1.0,
|
||||
"max_distance linear": 0.98,
|
||||
"max_distance rbf": 1.0,
|
||||
"max_distance poly": 1.0,
|
||||
"max_samples liblinear": 0.9550561797752809,
|
||||
"max_samples linear": 1.0,
|
||||
"max_samples rbf": 0.6685393258426966,
|
||||
"max_samples poly": 0.6853932584269663,
|
||||
"max_samples sigmoid": 0.6404494382022472,
|
||||
"impurity liblinear": 0.9550561797752809,
|
||||
"impurity linear": 1.0,
|
||||
"impurity rbf": 0.6685393258426966,
|
||||
"impurity poly": 0.6853932584269663,
|
||||
"impurity sigmoid": 0.6404494382022472,
|
||||
},
|
||||
}
|
||||
|
||||
for name, dataset in datasets.items():
|
||||
px, py = dataset
|
||||
for criteria in ["max_samples", "min_distance", "max_distance"]:
|
||||
for criteria in ["max_samples", "impurity"]:
|
||||
for kernel in self._kernels:
|
||||
clf = Stree(
|
||||
C=1e4,
|
||||
max_iter=1e4,
|
||||
multiclass_strategy="ovr"
|
||||
if kernel == "liblinear"
|
||||
else "ovo",
|
||||
kernel=kernel,
|
||||
random_state=self._random_state,
|
||||
)
|
||||
clf.fit(px, py)
|
||||
outcome = outcomes[name][f"{criteria} {kernel}"]
|
||||
self.assertAlmostEqual(outcome, clf.score(px, py))
|
||||
# print(f'"{criteria} {kernel}": {clf.score(px, py)},')
|
||||
self.assertAlmostEqual(
|
||||
outcome,
|
||||
clf.score(px, py),
|
||||
5,
|
||||
f"{name} - {criteria} - {kernel}",
|
||||
)
|
||||
|
||||
def test_max_features(self):
|
||||
n_features = 16
|
||||
@@ -251,6 +313,12 @@ class Stree_test(unittest.TestCase):
|
||||
with self.assertRaises(ValueError):
|
||||
_ = clf._initialize_max_features()
|
||||
|
||||
def test_wrong_max_features(self):
|
||||
X, y = load_dataset(n_features=15)
|
||||
clf = Stree(max_features=16)
|
||||
with self.assertRaises(ValueError):
|
||||
clf.fit(X, y)
|
||||
|
||||
def test_get_subspaces(self):
|
||||
dataset = np.random.random((10, 16))
|
||||
y = np.random.randint(0, 2, 10)
|
||||
@@ -288,16 +356,21 @@ class Stree_test(unittest.TestCase):
|
||||
clf.predict(X[:, :3])
|
||||
|
||||
# Tests of score
|
||||
|
||||
def test_score_binary(self):
|
||||
X, y = load_dataset(self._random_state)
|
||||
accuracies = [
|
||||
0.9506666666666667,
|
||||
0.9493333333333334,
|
||||
0.9606666666666667,
|
||||
0.9433333333333334,
|
||||
0.9153333333333333,
|
||||
]
|
||||
for kernel, accuracy_expected in zip(self._kernels, accuracies):
|
||||
clf = Stree(random_state=self._random_state, kernel=kernel,)
|
||||
clf = Stree(
|
||||
random_state=self._random_state,
|
||||
multiclass_strategy="ovr" if kernel == "liblinear" else "ovo",
|
||||
kernel=kernel,
|
||||
)
|
||||
clf.fit(X, y)
|
||||
accuracy_score = clf.score(X, y)
|
||||
yp = clf.predict(X)
|
||||
@@ -307,110 +380,284 @@ class Stree_test(unittest.TestCase):
|
||||
|
||||
def test_score_max_features(self):
|
||||
X, y = load_dataset(self._random_state)
|
||||
clf = Stree(random_state=self._random_state, max_features=2)
|
||||
clf = Stree(
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
random_state=self._random_state,
|
||||
max_features=2,
|
||||
)
|
||||
clf.fit(X, y)
|
||||
self.assertAlmostEqual(0.9426666666666667, clf.score(X, y))
|
||||
|
||||
def test_score_multi_class(self):
|
||||
warnings.filterwarnings("ignore")
|
||||
accuracies = [
|
||||
0.8258427, # Wine linear min_distance
|
||||
0.6741573, # Wine linear max_distance
|
||||
0.8314607, # Wine linear max_samples
|
||||
0.6629213, # Wine rbf min_distance
|
||||
1.0000000, # Wine rbf max_distance
|
||||
0.4044944, # Wine rbf max_samples
|
||||
0.9157303, # Wine poly min_distance
|
||||
1.0000000, # Wine poly max_distance
|
||||
0.7640449, # Wine poly max_samples
|
||||
0.9933333, # Iris linear min_distance
|
||||
0.9666667, # Iris linear max_distance
|
||||
0.9666667, # Iris linear max_samples
|
||||
0.9800000, # Iris rbf min_distance
|
||||
0.9800000, # Iris rbf max_distance
|
||||
0.9800000, # Iris rbf max_samples
|
||||
1.0000000, # Iris poly min_distance
|
||||
1.0000000, # Iris poly max_distance
|
||||
1.0000000, # Iris poly max_samples
|
||||
0.8993333, # Synthetic linear min_distance
|
||||
0.6533333, # Synthetic linear max_distance
|
||||
0.9313333, # Synthetic linear max_samples
|
||||
0.8320000, # Synthetic rbf min_distance
|
||||
0.6660000, # Synthetic rbf max_distance
|
||||
0.8320000, # Synthetic rbf max_samples
|
||||
0.6066667, # Synthetic poly min_distance
|
||||
0.6840000, # Synthetic poly max_distance
|
||||
0.6340000, # Synthetic poly max_samples
|
||||
]
|
||||
datasets = [
|
||||
("Wine", load_wine(return_X_y=True)),
|
||||
("Iris", load_iris(return_X_y=True)),
|
||||
(
|
||||
"Synthetic",
|
||||
load_dataset(self._random_state, n_classes=3, n_features=5),
|
||||
),
|
||||
]
|
||||
for dataset_name, dataset in datasets:
|
||||
X, y = dataset
|
||||
for kernel in self._kernels:
|
||||
for criteria in [
|
||||
"min_distance",
|
||||
"max_distance",
|
||||
"max_samples",
|
||||
]:
|
||||
clf = Stree(
|
||||
C=17,
|
||||
random_state=self._random_state,
|
||||
kernel=kernel,
|
||||
split_criteria=criteria,
|
||||
degree=5,
|
||||
gamma="auto",
|
||||
)
|
||||
clf.fit(X, y)
|
||||
accuracy_score = clf.score(X, y)
|
||||
yp = clf.predict(X)
|
||||
accuracy_computed = np.mean(yp == y)
|
||||
# print(
|
||||
# "{:.7f}, # {:7} {:5} {}".format(
|
||||
# accuracy_score, dataset_name, kernel, criteria
|
||||
# )
|
||||
# )
|
||||
accuracy_expected = accuracies.pop(0)
|
||||
self.assertEqual(accuracy_score, accuracy_computed)
|
||||
self.assertAlmostEqual(accuracy_expected, accuracy_score)
|
||||
self.assertAlmostEqual(0.9453333333333334, clf.score(X, y))
|
||||
|
||||
def test_bogus_splitter_parameter(self):
|
||||
clf = Stree(splitter="duck")
|
||||
with self.assertRaises(ValueError):
|
||||
clf.fit(*load_dataset())
|
||||
|
||||
def test_weights_removing_class(self):
|
||||
# This patch solves an stderr message from sklearn svm lib
|
||||
# "WARNING: class label x specified in weight is not found"
|
||||
def test_multiclass_classifier_integrity(self):
|
||||
"""Checks if the multiclass operation is done right"""
|
||||
X, y = load_iris(return_X_y=True)
|
||||
clf = Stree(
|
||||
kernel="liblinear", multiclass_strategy="ovr", random_state=0
|
||||
)
|
||||
clf.fit(X, y)
|
||||
score = clf.score(X, y)
|
||||
# Check accuracy of the whole model
|
||||
self.assertAlmostEquals(0.98, score, 5)
|
||||
svm = LinearSVC(random_state=0)
|
||||
svm.fit(X, y)
|
||||
self.assertAlmostEquals(0.9666666666666667, svm.score(X, y), 5)
|
||||
data = svm.decision_function(X)
|
||||
expected = [
|
||||
0.4444444444444444,
|
||||
0.35777777777777775,
|
||||
0.4569777777777778,
|
||||
]
|
||||
ty = data.copy()
|
||||
ty[data <= 0] = 0
|
||||
ty[data > 0] = 1
|
||||
ty = ty.astype(int)
|
||||
for i in range(3):
|
||||
self.assertAlmostEquals(
|
||||
expected[i],
|
||||
clf.splitter_._gini(ty[:, i]),
|
||||
)
|
||||
# 1st Branch
|
||||
# up has to have 50 samples of class 0
|
||||
# down should have 100 [50, 50]
|
||||
up = data[:, 2] > 0
|
||||
resup = np.unique(y[up], return_counts=True)
|
||||
resdn = np.unique(y[~up], return_counts=True)
|
||||
self.assertListEqual([1, 2], resup[0].tolist())
|
||||
self.assertListEqual([3, 50], resup[1].tolist())
|
||||
self.assertListEqual([0, 1], resdn[0].tolist())
|
||||
self.assertListEqual([50, 47], resdn[1].tolist())
|
||||
# 2nd Branch
|
||||
# up should have 53 samples of classes [1, 2] [3, 50]
|
||||
# down shoud have 47 samples of class 1
|
||||
node_up = clf.tree_.get_down().get_up()
|
||||
node_dn = clf.tree_.get_down().get_down()
|
||||
resup = np.unique(node_up._y, return_counts=True)
|
||||
resdn = np.unique(node_dn._y, return_counts=True)
|
||||
self.assertListEqual([1, 2], resup[0].tolist())
|
||||
self.assertListEqual([3, 50], resup[1].tolist())
|
||||
self.assertListEqual([1], resdn[0].tolist())
|
||||
self.assertListEqual([47], resdn[1].tolist())
|
||||
|
||||
def test_score_multiclass_rbf(self):
|
||||
X, y = load_dataset(
|
||||
random_state=self._random_state,
|
||||
n_classes=3,
|
||||
n_features=5,
|
||||
n_samples=500,
|
||||
)
|
||||
clf = Stree(kernel="rbf", random_state=self._random_state)
|
||||
clf2 = Stree(
|
||||
kernel="rbf", random_state=self._random_state, normalize=True
|
||||
)
|
||||
self.assertEqual(0.966, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(0.964, clf2.fit(X, y).score(X, y))
|
||||
X, y = load_wine(return_X_y=True)
|
||||
self.assertEqual(0.6685393258426966, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(1.0, clf2.fit(X, y).score(X, y))
|
||||
|
||||
def test_score_multiclass_poly(self):
|
||||
X, y = load_dataset(
|
||||
random_state=self._random_state,
|
||||
n_classes=3,
|
||||
n_features=5,
|
||||
n_samples=500,
|
||||
)
|
||||
clf = Stree(
|
||||
kernel="poly", random_state=self._random_state, C=10, degree=5
|
||||
)
|
||||
clf2 = Stree(
|
||||
kernel="poly",
|
||||
random_state=self._random_state,
|
||||
normalize=True,
|
||||
)
|
||||
self.assertEqual(0.946, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(0.972, clf2.fit(X, y).score(X, y))
|
||||
X, y = load_wine(return_X_y=True)
|
||||
self.assertEqual(0.7808988764044944, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(1.0, clf2.fit(X, y).score(X, y))
|
||||
|
||||
def test_score_multiclass_liblinear(self):
|
||||
X, y = load_dataset(
|
||||
random_state=self._random_state,
|
||||
n_classes=3,
|
||||
n_features=5,
|
||||
n_samples=500,
|
||||
)
|
||||
clf = Stree(
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
random_state=self._random_state,
|
||||
C=10,
|
||||
)
|
||||
clf2 = Stree(
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
random_state=self._random_state,
|
||||
normalize=True,
|
||||
)
|
||||
self.assertEqual(0.968, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(0.97, clf2.fit(X, y).score(X, y))
|
||||
X, y = load_wine(return_X_y=True)
|
||||
self.assertEqual(1.0, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(1.0, clf2.fit(X, y).score(X, y))
|
||||
|
||||
def test_score_multiclass_sigmoid(self):
|
||||
X, y = load_dataset(
|
||||
random_state=self._random_state,
|
||||
n_classes=3,
|
||||
n_features=5,
|
||||
n_samples=500,
|
||||
)
|
||||
clf = Stree(kernel="sigmoid", random_state=self._random_state, C=10)
|
||||
clf2 = Stree(
|
||||
kernel="sigmoid",
|
||||
random_state=self._random_state,
|
||||
normalize=True,
|
||||
C=10,
|
||||
)
|
||||
self.assertEqual(0.796, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(0.952, clf2.fit(X, y).score(X, y))
|
||||
X, y = load_wine(return_X_y=True)
|
||||
self.assertEqual(0.6910112359550562, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(0.9662921348314607, clf2.fit(X, y).score(X, y))
|
||||
|
||||
def test_score_multiclass_linear(self):
|
||||
warnings.filterwarnings("ignore", category=ConvergenceWarning)
|
||||
warnings.filterwarnings("ignore", category=RuntimeWarning)
|
||||
X, y = load_dataset(
|
||||
random_state=self._random_state,
|
||||
n_classes=3,
|
||||
n_features=5,
|
||||
n_samples=1500,
|
||||
)
|
||||
clf = Stree(
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
random_state=self._random_state,
|
||||
)
|
||||
self.assertEqual(0.9533333333333334, clf.fit(X, y).score(X, y))
|
||||
# Check with context based standardization
|
||||
clf2 = Stree(
|
||||
kernel="liblinear",
|
||||
multiclass_strategy="ovr",
|
||||
random_state=self._random_state,
|
||||
normalize=True,
|
||||
)
|
||||
self.assertEqual(0.9526666666666667, clf2.fit(X, y).score(X, y))
|
||||
X, y = load_wine(return_X_y=True)
|
||||
self.assertEqual(0.9831460674157303, clf.fit(X, y).score(X, y))
|
||||
self.assertEqual(1.0, clf2.fit(X, y).score(X, y))
|
||||
|
||||
def test_zero_all_sample_weights(self):
|
||||
X, y = load_dataset(self._random_state)
|
||||
with self.assertRaises(ValueError):
|
||||
Stree().fit(X, y, np.zeros(len(y)))
|
||||
|
||||
def test_mask_samples_weighted_zero(self):
|
||||
X = np.array(
|
||||
[
|
||||
[0.1, 0.1],
|
||||
[0.1, 0.2],
|
||||
[0.2, 0.1],
|
||||
[5, 6],
|
||||
[8, 9],
|
||||
[6, 7],
|
||||
[0.2, 0.2],
|
||||
[1, 1],
|
||||
[1, 1],
|
||||
[1, 1],
|
||||
[2, 2],
|
||||
[2, 2],
|
||||
[2, 2],
|
||||
[3, 3],
|
||||
[3, 3],
|
||||
[3, 3],
|
||||
]
|
||||
)
|
||||
y = np.array([0, 0, 0, 1, 1, 1, 0])
|
||||
epsilon = 1e-5
|
||||
weights = [1, 1, 1, 0, 0, 0, 1]
|
||||
weights = np.array(weights, dtype="float64")
|
||||
weights_epsilon = [x + epsilon for x in weights]
|
||||
weights_no_zero = np.array([1, 1, 1, 0, 0, 2, 1])
|
||||
original = weights_no_zero.copy()
|
||||
clf = Stree()
|
||||
y = np.array([1, 1, 1, 2, 2, 2, 5, 5, 5])
|
||||
yw = np.array([1, 1, 1, 1, 1, 1, 5, 5, 5])
|
||||
w = [1, 1, 1, 0, 0, 0, 1, 1, 1]
|
||||
model1 = Stree().fit(X, y)
|
||||
model2 = Stree().fit(X, y, w)
|
||||
predict1 = model1.predict(X)
|
||||
predict2 = model2.predict(X)
|
||||
self.assertListEqual(y.tolist(), predict1.tolist())
|
||||
self.assertListEqual(yw.tolist(), predict2.tolist())
|
||||
self.assertEqual(model1.score(X, y), 1)
|
||||
self.assertAlmostEqual(model2.score(X, y), 0.66666667)
|
||||
self.assertEqual(model2.score(X, y, w), 1)
|
||||
|
||||
def test_depth(self):
|
||||
X, y = load_dataset(
|
||||
random_state=self._random_state,
|
||||
n_classes=3,
|
||||
n_features=5,
|
||||
n_samples=1500,
|
||||
)
|
||||
clf = Stree(random_state=self._random_state)
|
||||
clf.fit(X, y)
|
||||
node = clf.train(X, y, weights, 1, "test",)
|
||||
# if a class is lost with zero weights the patch adds epsilon
|
||||
self.assertListEqual(weights.tolist(), weights_epsilon)
|
||||
self.assertListEqual(node._sample_weight.tolist(), weights_epsilon)
|
||||
# zero weights are ok when they don't erase a class
|
||||
_ = clf.train(X, y, weights_no_zero, 1, "test")
|
||||
self.assertListEqual(weights_no_zero.tolist(), original.tolist())
|
||||
self.assertEqual(6, clf.depth_)
|
||||
X, y = load_wine(return_X_y=True)
|
||||
clf = Stree(random_state=self._random_state)
|
||||
clf.fit(X, y)
|
||||
self.assertEqual(4, clf.depth_)
|
||||
|
||||
def test_nodes_leaves(self):
|
||||
X, y = load_dataset(
|
||||
random_state=self._random_state,
|
||||
n_classes=3,
|
||||
n_features=5,
|
||||
n_samples=1500,
|
||||
)
|
||||
clf = Stree(random_state=self._random_state)
|
||||
clf.fit(X, y)
|
||||
nodes, leaves = clf.nodes_leaves()
|
||||
self.assertEqual(31, nodes)
|
||||
self.assertEqual(16, leaves)
|
||||
X, y = load_wine(return_X_y=True)
|
||||
clf = Stree(random_state=self._random_state)
|
||||
clf.fit(X, y)
|
||||
nodes, leaves = clf.nodes_leaves()
|
||||
self.assertEqual(11, nodes)
|
||||
self.assertEqual(6, leaves)
|
||||
|
||||
def test_nodes_leaves_artificial(self):
|
||||
n1 = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test1")
|
||||
n2 = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test2")
|
||||
n3 = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test3")
|
||||
n4 = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test4")
|
||||
n5 = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test5")
|
||||
n6 = Snode(None, [1, 2, 3, 4], [1, 0, 1, 1], [], 0.0, "test6")
|
||||
n1.set_up(n2)
|
||||
n2.set_up(n3)
|
||||
n2.set_down(n4)
|
||||
n3.set_up(n5)
|
||||
n4.set_down(n6)
|
||||
clf = Stree(random_state=self._random_state)
|
||||
clf.tree_ = n1
|
||||
nodes, leaves = clf.nodes_leaves()
|
||||
self.assertEqual(6, nodes)
|
||||
self.assertEqual(2, leaves)
|
||||
|
||||
def test_bogus_multiclass_strategy(self):
|
||||
clf = Stree(multiclass_strategy="other")
|
||||
X, y = load_wine(return_X_y=True)
|
||||
with self.assertRaises(ValueError):
|
||||
clf.fit(X, y)
|
||||
|
||||
def test_multiclass_strategy(self):
|
||||
X, y = load_wine(return_X_y=True)
|
||||
clf_o = Stree(multiclass_strategy="ovo")
|
||||
clf_r = Stree(multiclass_strategy="ovr")
|
||||
score_o = clf_o.fit(X, y).score(X, y)
|
||||
score_r = clf_r.fit(X, y).score(X, y)
|
||||
self.assertEqual(1.0, score_o)
|
||||
self.assertEqual(0.9269662921348315, score_r)
|
||||
|
||||
def test_incompatible_hyperparameters(self):
|
||||
X, y = load_wine(return_X_y=True)
|
||||
clf = Stree(kernel="liblinear", multiclass_strategy="ovo")
|
||||
with self.assertRaises(ValueError):
|
||||
clf.fit(X, y)
|
||||
clf = Stree(multiclass_strategy="ovo", split_criteria="max_samples")
|
||||
with self.assertRaises(ValueError):
|
||||
clf.fit(X, y)
|
||||
|
||||
@@ -1,11 +1,14 @@
|
||||
from sklearn.datasets import make_classification
|
||||
import numpy as np
|
||||
|
||||
|
||||
def load_dataset(random_state=0, n_classes=2, n_features=3):
|
||||
def load_dataset(
|
||||
random_state=0, n_classes=2, n_features=3, n_samples=1500, n_informative=3
|
||||
):
|
||||
X, y = make_classification(
|
||||
n_samples=1500,
|
||||
n_samples=n_samples,
|
||||
n_features=n_features,
|
||||
n_informative=3,
|
||||
n_informative=n_informative,
|
||||
n_redundant=0,
|
||||
n_repeated=0,
|
||||
n_classes=n_classes,
|
||||
@@ -15,3 +18,12 @@ def load_dataset(random_state=0, n_classes=2, n_features=3):
|
||||
random_state=random_state,
|
||||
)
|
||||
return X, y
|
||||
|
||||
|
||||
def load_disc_dataset(
|
||||
random_state=0, n_classes=2, n_features=3, n_samples=1500
|
||||
):
|
||||
np.random.seed(random_state)
|
||||
X = np.random.randint(1, 17, size=(n_samples, n_features)).astype(float)
|
||||
y = np.random.randint(low=0, high=n_classes, size=(n_samples), dtype=int)
|
||||
return X, y
|
||||
|
||||
Reference in New Issue
Block a user