bugfix for predict with gamma = 'auto'

4 files changed, 26 insertions, 2 deletions

diff --git a/CHANGELOG.rst b/CHANGELOG.rst
index fd5bedb..be4d1df 100644
--- a/CHANGELOG.rst
+++ b/CHANGELOG.rst
@@ -1,6 +1,11 @@
 Change Log
 ----------
 
+Version 0.2.3
+^^^^^^^^^^^^^
+
+- Bugfix for prediction with gamma = 'auto'
+
 Version 0.2.2
 ^^^^^^^^^^^^^
 
diff --git a/gensvm/__init__.py b/gensvm/__init__.py
index 62ac067..c8cd0be 100644
--- a/gensvm/__init__.py
+++ b/gensvm/__init__.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 
-__version__ = "0.2.2"
+__version__ = "0.2.3"
 
 from .core import GenSVM
 from .gridsearch import GenSVMGridSearchCV
diff --git a/gensvm/core.py b/gensvm/core.py
index c994995..f0e7820 100644
--- a/gensvm/core.py
+++ b/gensvm/core.py
@@ -408,6 +408,8 @@ class GenSVM(BaseEstimator, ClassifierMixin):
                 trainX, accept_sparse=False, dtype=np.float64, order="C"
             )
 
+        gamma = 1.0 / X.shape[1] if self.gamma == "auto" else self.gamma
+
         V = self.combined_coef_
         if self.kernel == "linear":
             predictions = wrapper.predict_wrap(X, V)
@@ -420,7 +422,7 @@ class GenSVM(BaseEstimator, ClassifierMixin):
                 V,
                 n_class,
                 kernel_idx,
-                self.gamma,
+                gamma,
                 self.coef,
                 self.degree,
                 self.kernel_eigen_cutoff,
diff --git a/test/test_core.py b/test/test_core.py
index 0a1042a..63c172d 100644
--- a/test/test_core.py
+++ b/test/test_core.py
@@ -197,6 +197,23 @@ class GenSVMTestCase(unittest.TestCase):
             set(pred).issubset(set(["versicolor", "virginica", "setosa"]))
         )
 
+    def test_fit_nonlinear_auto(self):
+        """ GENSVM: Fit and predict with nonlinear kernel """
+        data = load_iris()
+        X = data.data
+        y = data.target_names[data.target]
+
+        X_train, X_test, y_train, y_test = train_test_split(
+            X, y, random_state=123
+        )
+        clf = GenSVM(kernel="rbf", max_iter=1000, random_state=123)
+        clf.fit(X_train, y_train)
+
+        pred = clf.predict(X_test, trainX=X_train)
+        self.assertTrue(
+            set(pred).issubset(set(["versicolor", "virginica", "setosa"]))
+        )
+
     def test_fit_with_seed(self):
         """ GENSVM: Test fit with seeding """