from sklearn.svm import SVC, LinearSVC
from sklearn import metrics
from sklearn.cross_validation import StratifiedKFold
from sklearn.datasets import load_svmlight_file

if __name__ == "__main__":

    X, y = load_svmlight_file("ntcir.en.vec")

    kf = StratifiedKFold(y, k=5, indices=True)

    clf = SVC(kernel="linear")
    #clf = LinearSVC()
    for train_index, test_index in kf:

        y_train, y_test = y[train_index], y[test_index]

        X_train, X_test = X[train_index], X[test_index]
        clf.fit(X_train, y_train)

        y_predict = clf.predict(X_test)

        print metrics.confusion_matrix(y_test, y_predict)
        print metrics.classification_report(y_test, y_predict)