clf.fit(X_train, y_train)
predicted_y = clf.calculate_prediction(X_test)
svm_accuracy.append(accuracy_score(y_test,predicted_y))
svm_confusion_mat = confusion_matrix(y_test, predicted_y)
sv_accuracy, svm_precision_val, svm_recall_val, svm_f_measure_val = clf.svm_findOtherParameters(svm_confusion_mat)
# SVM end
# Naive Bayes
clf_NB = BernoulliNB(alpha=1.0, binarize=0.0, class_prior=None, fit_prior=True)
clf_NB.fit(X_train, y_train)
predictedBNB_y = clf_NB.predict(X_test)
NBSKL_accuracy.append(accuracy_score(y_test,predictedBNB_y))
nb = NaiveBayes.NaiveBayesBernoulli()
# iterate data for each class
for clas in np.unique(y):
class_feature_matrix = X_train[y_train==clas]
prior_array = len(class_feature_matrix)*1.0/len(X_train)
# print prior_array
alpha = [(np.sum(class_feature_matrix[:,i])/len(class_feature_matrix)) for i in range(class_feature_matrix.shape[1])]
gX = nb.membership_function(X_test, alpha, prior_array)
max_gX.update({int(clas): gX})
# find discriminant function
disc_function = nb.discriminant_function(max_gX, np.unique(y))
# print disc_function
confusion_mat = confusion_matrix(y_test, predictedBNB_y)
# print confusion_mat
