Merge pull request #5216 from JPFrancoia/master

agramfort · agramfort · commit 9ebc6fe33d85 · 2015-09-11T11:42:39.000+02:00
Creation of the attribute LDA.explained_variance_ratio_, for the eige…
diff --git a/sklearn/lda.py b/sklearn/lda.py
@@ -180,6 +180,12 @@ class LDA(BaseEstimator, LinearClassifierMixin, TransformerMixin):
     covariance_ : array-like, shape (n_features, n_features)
         Covariance matrix (shared by all classes).
 
+    explained_variance_ratio_ : array, shape (n_components,)
+        Percentage of variance explained by each of the selected components.
+        If ``n_components`` is not set then all components are stored and the
+        sum of explained variances is equal to 1.0. Only available when eigen
+        solver is used.
+
     means_ : array-like, shape (n_classes, n_features)
         Class means.
 
@@ -316,6 +322,7 @@ class scatter). This solver supports both classification and
         Sb = St - Sw  # between scatter
 
         evals, evecs = linalg.eigh(Sb, Sw)
+        self.explained_variance_ratio_ = np.sort(evals / np.sum(evals))[::-1]
         evecs = evecs[:, np.argsort(evals)[::-1]]  # sort eigenvectors
         # evecs /= np.linalg.norm(evecs, axis=0)  # doesn't work with numpy 1.6
         evecs /= np.apply_along_axis(np.linalg.norm, 0, evecs)
diff --git a/sklearn/tests/test_lda.py b/sklearn/tests/test_lda.py
@@ -85,6 +85,21 @@ def test_lda_coefs():
     assert_array_almost_equal(clf_lda_eigen.coef_, clf_lda_lsqr.coef_, 1)
 
 
+def test_lda_explained_variance_ratio():
+    # Test if the sum of the normalized eigen vectors values equals 1
+    n_features = 2
+    n_classes = 2
+    n_samples = 1000
+    X, y = make_blobs(n_samples=n_samples, n_features=n_features,
+                      centers=n_classes, random_state=11)
+
+    clf_lda_eigen = lda.LDA(solver="eigen")
+
+    clf_lda_eigen.fit(X, y)
+
+    assert_almost_equal(clf_lda_eigen.explained_variance_ratio_.sum(), 1.0, 3)
+
+
 def test_lda_transform():
     # Test LDA transform.
     clf = lda.LDA(solver="svd", n_components=1)
