Add nodes, leaves, depth to mysql

Add nodes, leaves, depth, samples, features and classes to analysis

experimentation/Database.py

@@ -176,6 +176,7 @@ class BD(ABC):
         accuracy,
         time_spent,
         parameters,
+        complexity,
     ) -> None:
         """Create a record in MySQL database
 
@@ -187,8 +188,8 @@ class BD(ABC):
         command_insert = (
             "replace into results (date, time, type, accuracy, "
             "dataset, classifier, norm, stand, parameters, accuracy_std, "
-            "time_spent, time_spent_std) values (%s, %s, "
-            "%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)"
+            "time_spent, time_spent_std, nodes, leaves, depth) values (%s, %s,"
+            " %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)"
         )
         now = datetime.now()
         date = now.strftime("%Y-%m-%d")
@@ -206,6 +207,9 @@ class BD(ABC):
             accuracy[1],
             time_spent[0],
             time_spent[1],
+            complexity["nodes"],
+            complexity["leaves"],
+            complexity["depth"],
         )
         cursor = database.cursor()
         cursor.execute(command_insert, values)
@@ -319,7 +323,9 @@ class Outcomes(BD):
         self._table = "outcomes"
         super().__init__(host=host, model=model)
 
-    def store(self, dataset, normalize, standardize, parameters, results):
+    def store(
+        self, dataset, normalize, standardize, parameters, results, complexity
+    ):
         outcomes = ["fit_time", "score_time", "train_score", "test_score"]
         data = ""
         for index in outcomes:
@@ -350,6 +356,7 @@ class Outcomes(BD):
                 float(np.std(results["fit_time"])),
             ],
             parameters,
+            complexity,
         )
 
     def report(self, dataset, exclude_params):

experimentation/Experiments.py

@@ -2,9 +2,8 @@ import json
 import os
 import time
 import warnings
-
+import numpy as np
 from sklearn.model_selection import GridSearchCV, cross_validate
-
 from . import Models
 from .Database import Hyperparameters, MySQL, Outcomes
 from .Sets import Datasets
@@ -94,15 +93,25 @@ class Experiment:
                     X,
                     y,
                     return_train_score=True,
+                    return_estimator=True,
                     n_jobs=self._threads,
                     cv=kfold,
                 )
                 for item in outcomes:
                     total[item].append(results[item])
         print("end")
+        if type(model).__name__ == "Stree":
+            best_model = results["estimator"][np.argmax(results["test_score"])]
+            nodes, leaves = best_model.nodes_leaves()
+            depth = best_model.depth_
+        else:
+            nodes = leaves = depth = 0
+        complexity = dict(nodes=nodes, leaves=leaves, depth=depth)
         outcomes = Outcomes(host=self._host, model=self._model_name)
         parameters = json.dumps(parameters, sort_keys=True)
-        outcomes.store(dataset, normalize, standardize, parameters, total)
+        outcomes.store(
+            dataset, normalize, standardize, parameters, total, complexity
+        )
         if self._num_warnings > 0:
             print(f"{self._num_warnings} warnings have happend")