ave.py 文件源码

def get_best_C(dataset):
    """
    Find the best parameters on the dev set.
    """
    best_f1 = 0
    best_c = 0

    labels = sorted(set(dataset._ytrain))

    test_cs = [0.001, 0.003, 0.006, 0.009,
                   0.01,  0.03,  0.06,  0.09,
                   0.1,   0.3,   0.6,   0.9,
                   1,       3,    6,     9,
                   10,      30,   60,    90]
    for i, c in enumerate(test_cs):

        sys.stdout.write('\rRunning cross-validation: {0} of {1}'.format(i+1, len(test_cs)))
        sys.stdout.flush()

        clf = LogisticRegression(C=c)
        h = clf.fit(dataset._Xtrain, dataset._ytrain)
        pred = clf.predict(dataset._Xdev)
        if len(labels) == 2:
            dev_f1 = f1_score(dataset._ydev, pred, pos_label=1)
        else:
            dev_f1 = f1_score(dataset._ydev, pred, labels=labels, average='micro')
        if dev_f1 > best_f1:
            best_f1 = dev_f1
            best_c = c

    print()
    print('Best F1 on dev data: {0:.3f}'.format(best_f1))
    print('Best C on dev data: {0}'.format(best_c))

    return best_c, best_f1