added margin as well

ISLP/svm.py

@@ -28,6 +28,12 @@ def plot(X,
    '''
    Graphical representation of fitted support vector classifier.
 
+   There are two types of support vectors:
+
+       - Points violating the margin but correctly classified. These are marked with a black '+'.
+
+       - Misclassified points. These are marked with a red 'x'.
+
    Parameters
    ----------
 
@@ -113,9 +119,19 @@ def plot(X,
                cmap=decision_cmap,
                alpha=alpha)
 
+   decision_val = svm.decision_function(X_pred)
+
    # add the support vectors    
 
    if svm.classes_.shape[0] == 2: # 2-class problem
+
+      ax.contourf(xval,
+                  yval,
+                  decision_val.reshape(yval.shape),
+                  levels=[-1,1],
+                  cmap=decision_cmap,
+                  alpha=alpha)
+
       D = svm.decision_function(X[svm.support_])
       Y_ = (2 * (Y[svm.support_] == svm.classes_[1]) - 1)
       violate_margin = (Y_ * D) > 0