naive_bayes.py 文件源码

def main():
    args = get_args()
    # load and split data
    dataset, target = load_dataset(args.file)
    train_x, train_y, test_x, actual = split_dataset(
        dataset, target, args.split)
    print("Training set size: %d, Testing set size: %d" %
          (len(train_x), len(test_x)))
    # prepare model
    summaries = summarize_by_class(train_x, train_y)
    # test model
    predictions = get_predictions(summaries, test_x)
    display(actual, predictions)
    # using scikit
    gnb = GaussianNB()
    y_pred = gnb.fit(train_x, train_y).predict(test_x)
    display(actual, y_pred)