Changed SVM to use gamma='auto' by default. Deprecated gamma=0.0.

Author: Shivan Sornarajah

doc/modules/model_persistence.rst

@@ -22,9 +22,9 @@ persistence model, namely `pickle <http://docs.python.org/library/pickle.html>`_
   >>> iris = datasets.load_iris()
   >>> X, y = iris.data, iris.target
   >>> clf.fit(X, y)  # doctest: +NORMALIZE_WHITESPACE
-  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,
-    kernel='rbf', max_iter=-1, probability=False, random_state=None,
-    shrinking=True, tol=0.001, verbose=False)
+  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, 
+    gamma='auto', kernel='rbf', max_iter=-1, probability=False, 
+    random_state=None, shrinking=True, tol=0.001, verbose=False)
 
   >>> import pickle
   >>> s = pickle.dumps(clf)

doc/modules/pipeline.rst

@@ -42,7 +42,7 @@ is an estimator object::
     >>> clf # doctest: +NORMALIZE_WHITESPACE
     Pipeline(steps=[('reduce_dim', PCA(copy=True, n_components=None,
         whiten=False)), ('svm', SVC(C=1.0, cache_size=200, class_weight=None,
-        coef0=0.0, degree=3, gamma=0.0, kernel='rbf', max_iter=-1,
+        coef0=0.0, degree=3, gamma='auto', kernel='rbf', max_iter=-1,
         probability=False, random_state=None, shrinking=True, tol=0.001,
         verbose=False))])
 
@@ -76,7 +76,7 @@ Parameters of the estimators in the pipeline can be accessed using the
     >>> clf.set_params(svm__C=10) # doctest: +NORMALIZE_WHITESPACE
     Pipeline(steps=[('reduce_dim', PCA(copy=True, n_components=None,
         whiten=False)), ('svm', SVC(C=10, cache_size=200, class_weight=None,
-        coef0=0.0, degree=3, gamma=0.0, kernel='rbf', max_iter=-1,
+        coef0=0.0, degree=3, gamma='auto', kernel='rbf', max_iter=-1,
         probability=False, random_state=None, shrinking=True, tol=0.001,
         verbose=False))])
 

doc/modules/svm.rst

@@ -77,8 +77,8 @@ n_features]`` holding the training samples, and an array y of class labels
     >>> clf = svm.SVC()
     >>> clf.fit(X, y)  # doctest: +NORMALIZE_WHITESPACE
     SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
-    gamma=0.0, kernel='rbf', max_iter=-1, probability=False, random_state=None,
-    shrinking=True, tol=0.001, verbose=False)
+    gamma='auto', kernel='rbf', max_iter=-1, probability=False, 
+    random_state=None, shrinking=True, tol=0.001, verbose=False)
 
 After being fitted, the model can then be used to predict new values::
 
@@ -116,8 +116,8 @@ classifiers are constructed and each one trains data from two classes::
     >>> clf = svm.SVC()
     >>> clf.fit(X, Y) # doctest: +NORMALIZE_WHITESPACE
     SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
-    gamma=0.0, kernel='rbf', max_iter=-1, probability=False, random_state=None,
-    shrinking=True, tol=0.001, verbose=False)
+    gamma='auto', kernel='rbf', max_iter=-1, probability=False, 
+    random_state=None, shrinking=True, tol=0.001, verbose=False)
     >>> dec = clf.decision_function([[1]])
     >>> dec.shape[1] # 4 classes: 4*3/2 = 6
     6
@@ -304,7 +304,7 @@ floating point values instead of integer values::
     >>> y = [0.5, 2.5]
     >>> clf = svm.SVR()
     >>> clf.fit(X, y) # doctest: +NORMALIZE_WHITESPACE
-    SVR(C=1.0, cache_size=200, coef0=0.0, degree=3, epsilon=0.1, gamma=0.0,
+    SVR(C=1.0, cache_size=200, coef0=0.0, degree=3, epsilon=0.1, gamma='auto',
         kernel='rbf', max_iter=-1, shrinking=True, tol=0.001, verbose=False)
     >>> clf.predict([[1, 1]])
     array([ 1.5])
@@ -504,7 +504,7 @@ test vectors must be provided.
     >>> gram = np.dot(X, X.T)
     >>> clf.fit(gram, y) # doctest: +NORMALIZE_WHITESPACE
     SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
-    gamma=0.0, kernel='precomputed', max_iter=-1, probability=False,
+    gamma='auto', kernel='precomputed', max_iter=-1, probability=False,
     random_state=None, shrinking=True, tol=0.001, verbose=False)
     >>> # predict on training examples
     >>> clf.predict(gram)

doc/tutorial/basic/tutorial.rst

@@ -214,9 +214,9 @@ persistence model, namely `pickle <http://docs.python.org/library/pickle.html>`_
   >>> iris = datasets.load_iris()
   >>> X, y = iris.data, iris.target
   >>> clf.fit(X, y)  # doctest: +NORMALIZE_WHITESPACE
-  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,
-    kernel='rbf', max_iter=-1, probability=False, random_state=None,
-    shrinking=True, tol=0.001, verbose=False)
+  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
+    gamma='auto', kernel='rbf', max_iter=-1, probability=False,
+    random_state=None, shrinking=True, tol=0.001, verbose=False)
 
   >>> import pickle
   >>> s = pickle.dumps(clf)
@@ -287,17 +287,17 @@ maintained::
     >>> iris = datasets.load_iris()
     >>> clf = SVC()
     >>> clf.fit(iris.data, iris.target)
-    SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,
-      kernel='rbf', max_iter=-1, probability=False, random_state=None,
-      shrinking=True, tol=0.001, verbose=False)
+    SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
+      gamma='auto', kernel='rbf', max_iter=-1, probability=False,
+      random_state=None, shrinking=True, tol=0.001, verbose=False)
 
     >>> list(clf.predict(iris.data[:3]))
     [0, 0, 0]
 
     >>> clf.fit(iris.data, iris.target_names[iris.target])
-    SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,
-      kernel='rbf', max_iter=-1, probability=False, random_state=None,
-      shrinking=True, tol=0.001, verbose=False)
+    SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
+      gamma='auto', kernel='rbf', max_iter=-1, probability=False,
+      random_state=None, shrinking=True, tol=0.001, verbose=False)
 
     >>> list(clf.predict(iris.data[:3]))  # doctest: +NORMALIZE_WHITESPACE
     ['setosa', 'setosa', 'setosa']
@@ -324,16 +324,16 @@ more than once will overwrite what was learned by any previous ``fit()``::
 
   >>> clf = SVC()
   >>> clf.set_params(kernel='linear').fit(X, y)
-  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,
-    kernel='linear', max_iter=-1, probability=False, random_state=None,
-    shrinking=True, tol=0.001, verbose=False)
+  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
+    gamma='auto', kernel='linear', max_iter=-1, probability=False,
+    random_state=None, shrinking=True, tol=0.001, verbose=False)
   >>> clf.predict(X_test)
   array([1, 0, 1, 1, 0])
 
   >>> clf.set_params(kernel='rbf').fit(X, y)
-  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,
-    kernel='rbf', max_iter=-1, probability=False, random_state=None,
-    shrinking=True, tol=0.001, verbose=False)
+  SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
+    gamma='auto', kernel='rbf', max_iter=-1, probability=False,
+    random_state=None, shrinking=True, tol=0.001, verbose=False)
   >>> clf.predict(X_test)
   array([0, 0, 0, 1, 0])
 

doc/tutorial/statistical_inference/supervised_learning.rst

@@ -453,9 +453,9 @@ classification --:class:`SVC` (Support Vector Classification).
     >>> from sklearn import svm
     >>> svc = svm.SVC(kernel='linear')
     >>> svc.fit(iris_X_train, iris_y_train)    # doctest: +NORMALIZE_WHITESPACE
-    SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, gamma=0.0,
-      kernel='linear', max_iter=-1, probability=False, random_state=None,
-      shrinking=True, tol=0.001, verbose=False)
+    SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3, 
+      gamma='auto', kernel='linear', max_iter=-1, probability=False, 
+      random_state=None, shrinking=True, tol=0.001, verbose=False)
 
 
 .. warning:: **Normalizing data**

sklearn/svm/base.py

@@ -72,6 +72,15 @@ def __init__(self, impl, kernel, degree, gamma, coef0,
         if impl not in LIBSVM_IMPL:  # pragma: no cover
             raise ValueError("impl should be one of %s, %s was given" % (
                 LIBSVM_IMPL, impl))
+       
+        # FIXME Remove gamma=0.0 support in 0.18 
+        if gamma == 0:
+            msg = ("gamma=%s has been deprecated in favor of "
+                   "gamma='%s' as of 0.17. Backward compatibility"
+                   " for gamma=%s will be removed in %s")
+            invalid_gamma = 0.0
+            warnings.warn(msg % (invalid_gamma, "auto", 
+                invalid_gamma, "0.18"), DeprecationWarning)
 
         self._impl = impl
         self.kernel = kernel
@@ -159,10 +168,14 @@ def fit(self, X, y, sample_weight=None):
                              "Note: Sparse matrices cannot be indexed w/"
                              "boolean masks (use `indices=True` in CV)."
                              % (sample_weight.shape, X.shape))
-
-        if (self.kernel in ['poly', 'rbf']) and (self.gamma == 0):
+        
+        # FIXME remove (self.gamma == 0) in 0.18
+        if (self.kernel in ['poly', 'rbf']) and ((self.gamma == 0) 
+                or (self.gamma == 'auto')):
             # if custom gamma is not provided ...
             self._gamma = 1.0 / X.shape[1]
+        elif self.gamma == 'auto':
+            self._gamma = 0.0
         else:
             self._gamma = self.gamma
 

sklearn/svm/classes.py

@@ -411,9 +411,9 @@ class SVC(BaseSVC):
         Degree of the polynomial kernel function ('poly').
         Ignored by all other kernels.
 
-    gamma : float, optional (default=0.0)
+    gamma : float, optional (default='auto')
         Kernel coefficient for 'rbf', 'poly' and 'sigmoid'.
-        If gamma is 0.0 then 1/n_features will be used instead.
+        If gamma is 'auto' then 1/n_features will be used instead.
 
     coef0 : float, optional (default=0.0)
         Independent term in kernel function.
@@ -489,7 +489,7 @@ class SVC(BaseSVC):
     >>> clf = SVC()
     >>> clf.fit(X, y) #doctest: +NORMALIZE_WHITESPACE
     SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0, degree=3,
-        gamma=0.0, kernel='rbf', max_iter=-1, probability=False,
+        gamma='auto', kernel='rbf', max_iter=-1, probability=False,
         random_state=None, shrinking=True, tol=0.001, verbose=False)
     >>> print(clf.predict([[-0.8, -1]]))
     [1]
@@ -506,7 +506,7 @@ class SVC(BaseSVC):
 
     """
 
-    def __init__(self, C=1.0, kernel='rbf', degree=3, gamma=0.0,
+    def __init__(self, C=1.0, kernel='rbf', degree=3, gamma='auto',
                  coef0=0.0, shrinking=True, probability=False,
                  tol=1e-3, cache_size=200, class_weight=None,
                  verbose=False, max_iter=-1, random_state=None):
@@ -545,9 +545,9 @@ class NuSVC(BaseSVC):
         Degree of the polynomial kernel function ('poly').
         Ignored by all other kernels.
 
-    gamma : float, optional (default=0.0)
+    gamma : float, optional (default='auto')
         Kernel coefficient for 'rbf', 'poly' and 'sigmoid'.
-        If gamma is 0.0 then 1/n_features will be used instead.
+        If gamma is 'auto' then 1/n_features will be used instead.
 
     coef0 : float, optional (default=0.0)
         Independent term in kernel function.
@@ -614,7 +614,7 @@ class NuSVC(BaseSVC):
     >>> from sklearn.svm import NuSVC
     >>> clf = NuSVC()
     >>> clf.fit(X, y) #doctest: +NORMALIZE_WHITESPACE
-    NuSVC(cache_size=200, coef0=0.0, degree=3, gamma=0.0, kernel='rbf',
+    NuSVC(cache_size=200, coef0=0.0, degree=3, gamma='auto', kernel='rbf',
           max_iter=-1, nu=0.5, probability=False, random_state=None,
           shrinking=True, tol=0.001, verbose=False)
     >>> print(clf.predict([[-0.8, -1]]))
@@ -630,7 +630,7 @@ class NuSVC(BaseSVC):
         liblinear.
     """
 
-    def __init__(self, nu=0.5, kernel='rbf', degree=3, gamma=0.0,
+    def __init__(self, nu=0.5, kernel='rbf', degree=3, gamma='auto',
                  coef0=0.0, shrinking=True, probability=False,
                  tol=1e-3, cache_size=200, verbose=False, max_iter=-1,
                  random_state=None):
@@ -671,9 +671,9 @@ class SVR(BaseLibSVM, RegressorMixin):
         Degree of the polynomial kernel function ('poly').
         Ignored by all other kernels.
 
-    gamma : float, optional (default=0.0)
+    gamma : float, optional (default='auto')
         Kernel coefficient for 'rbf', 'poly' and 'sigmoid'.
-        If gamma is 0.0 then 1/n_features will be used instead.
+        If gamma is 'auto' then 1/n_features will be used instead.
 
     coef0 : float, optional (default=0.0)
         Independent term in kernel function.
@@ -727,7 +727,7 @@ class SVR(BaseLibSVM, RegressorMixin):
     >>> X = np.random.randn(n_samples, n_features)
     >>> clf = SVR(C=1.0, epsilon=0.2)
     >>> clf.fit(X, y) #doctest: +NORMALIZE_WHITESPACE
-    SVR(C=1.0, cache_size=200, coef0=0.0, degree=3, epsilon=0.2, gamma=0.0,
+    SVR(C=1.0, cache_size=200, coef0=0.0, degree=3, epsilon=0.2, gamma='auto',
         kernel='rbf', max_iter=-1, shrinking=True, tol=0.001, verbose=False)
 
     See also
@@ -740,9 +740,9 @@ class SVR(BaseLibSVM, RegressorMixin):
         Scalable Linear Support Vector Machine for regression
         implemented using liblinear.
     """
-    def __init__(self, kernel='rbf', degree=3, gamma=0.0, coef0=0.0, tol=1e-3,
-                 C=1.0, epsilon=0.1, shrinking=True, cache_size=200,
-                 verbose=False, max_iter=-1):
+    def __init__(self, kernel='rbf', degree=3, gamma='auto', coef0=0.0, 
+                 tol=1e-3, C=1.0, epsilon=0.1, shrinking=True, 
+                 cache_size=200, verbose=False, max_iter=-1):
 
         super(SVR, self).__init__(
             'epsilon_svr', kernel=kernel, degree=degree, gamma=gamma,
@@ -783,9 +783,9 @@ class NuSVR(BaseLibSVM, RegressorMixin):
         Degree of the polynomial kernel function ('poly').
         Ignored by all other kernels.
 
-    gamma : float, optional (default=0.0)
+    gamma : float, optional (default='auto')
         Kernel coefficient for 'rbf', 'poly' and 'sigmoid'.
-        If gamma is 0.0 then 1/n_features will be used instead.
+        If gamma is 'auto' then 1/n_features will be used instead.
 
     coef0 : float, optional (default=0.0)
         Independent term in kernel function.
@@ -839,8 +839,9 @@ class NuSVR(BaseLibSVM, RegressorMixin):
     >>> X = np.random.randn(n_samples, n_features)
     >>> clf = NuSVR(C=1.0, nu=0.1)
     >>> clf.fit(X, y)  #doctest: +NORMALIZE_WHITESPACE
-    NuSVR(C=1.0, cache_size=200, coef0=0.0, degree=3, gamma=0.0, kernel='rbf',
-          max_iter=-1, nu=0.1, shrinking=True, tol=0.001, verbose=False)
+    NuSVR(C=1.0, cache_size=200, coef0=0.0, degree=3, gamma='auto', 
+          kernel='rbf', max_iter=-1, nu=0.1, shrinking=True, tol=0.001, 
+          verbose=False)
 
     See also
     --------
@@ -853,7 +854,7 @@ class NuSVR(BaseLibSVM, RegressorMixin):
     """
 
     def __init__(self, nu=0.5, C=1.0, kernel='rbf', degree=3,
-                 gamma=0.0, coef0=0.0, shrinking=True, tol=1e-3,
+                 gamma='auto', coef0=0.0, shrinking=True, tol=1e-3,
                  cache_size=200, verbose=False, max_iter=-1):
 
         super(NuSVR, self).__init__(
@@ -891,9 +892,9 @@ class OneClassSVM(BaseLibSVM):
         Degree of the polynomial kernel function ('poly').
         Ignored by all other kernels.
 
-    gamma : float, optional (default=0.0)
+    gamma : float, optional (default='auto')
         Kernel coefficient for 'rbf', 'poly' and 'sigmoid'.
-        If gamma is 0.0 then 1/n_features will be used instead.
+        If gamma is 'auto' then 1/n_features will be used instead.
 
     coef0 : float, optional (default=0.0)
         Independent term in kernel function.
@@ -942,9 +943,9 @@ class OneClassSVM(BaseLibSVM):
         Constants in decision function.
 
     """
-    def __init__(self, kernel='rbf', degree=3, gamma=0.0, coef0=0.0, tol=1e-3,
-                 nu=0.5, shrinking=True, cache_size=200, verbose=False,
-                 max_iter=-1, random_state=None):
+    def __init__(self, kernel='rbf', degree=3, gamma='auto', coef0=0.0, 
+                 tol=1e-3, nu=0.5, shrinking=True, cache_size=200, 
+                 verbose=False, max_iter=-1, random_state=None):
 
         super(OneClassSVM, self).__init__(
             'one_class', kernel, degree, gamma, coef0, tol, 0., nu, 0.,