From 11b473d5605d6e8c9a2f9682d23b54ac59f9eced Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ricardo=20Montan=CC=83ana?= <rmontanana@gmail.com>
Date: Tue, 19 Jan 2021 11:13:23 +0100
Subject: [PATCH] Fix problem with zero weighted samples Solve WARNING: class
 label x specified in weight is not found with a different approach

---
 requirements.txt          |  2 +-
 setup.py                  |  4 +--
 stree/Strees.py           | 14 +++++-----
 stree/tests/Stree_test.py | 54 ++++++++++++++++-----------------------
 4 files changed, 31 insertions(+), 43 deletions(-)

diff --git a/requirements.txt b/requirements.txt
index 2359d77..dbfe578 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,4 @@
 numpy
-scikit-learn==0.23.2
+scikit-learn
 pandas
 ipympl
\ No newline at end of file
diff --git a/setup.py b/setup.py
index adf861d..588f7a3 100644
--- a/setup.py
+++ b/setup.py
@@ -1,6 +1,6 @@
 import setuptools
 
-__version__ = "0.9rc6"
+__version__ = "1.0rc1"
 __author__ = "Ricardo Montañana Gómez"
 
 
@@ -30,7 +30,7 @@ setuptools.setup(
         "Topic :: Scientific/Engineering :: Artificial Intelligence",
         "Intended Audience :: Science/Research",
     ],
-    install_requires=["scikit-learn==0.23.2", "numpy", "ipympl"],
+    install_requires=["scikit-learn", "numpy", "ipympl"],
     test_suite="stree.tests",
     zip_safe=False,
 )
diff --git a/stree/Strees.py b/stree/Strees.py
index 9bdd511..ab35b4b 100644
--- a/stree/Strees.py
+++ b/stree/Strees.py
@@ -629,6 +629,12 @@ class Stree(BaseEstimator, ClassifierMixin):
         """
         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]
         if np.unique(y).shape[0] == 1:
             # only 1 class => pure dataset
             return Snode(
@@ -643,14 +649,6 @@ class Stree(BaseEstimator, ClassifierMixin):
         # 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
         clf.fit(Xs, y, sample_weight=sample_weight)
         impurity = self.splitter_.partition_impurity(y)
         node = Snode(clf, X, y, features, impurity, title, sample_weight)
diff --git a/stree/tests/Stree_test.py b/stree/tests/Stree_test.py
index 4cd6b59..65afeef 100644
--- a/stree/tests/Stree_test.py
+++ b/stree/tests/Stree_test.py
@@ -413,39 +413,29 @@ class Stree_test(unittest.TestCase):
         with self.assertRaises(ValueError):
             Stree().fit(X, y, np.zeros(len(y)))
 
-    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_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()
-        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())
+        y = np.array([1, 1, 1, 2, 2, 2, 5, 5, 5])
+        yw = np.array([1, 1, 1, 5, 5, 5, 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)