solve.py 文件源码

def solve():
    # ????????????
    training_arr = numpy.loadtxt('adult.txt', dtype=bytes, comments='#', delimiter=',')
    test_data = numpy.loadtxt('adult_test.txt', dtype=bytes, comments='#', delimiter=',')

    # ??? x ???? y ???, ?????
    # x_list = numpy.ndarray(len(training_arr))
    # y_list = numpy.ndarray(len(training_arr))

    # ????? y ?
    y_list = [int(element[12]) for element in training_arr]
    # ????, ??????? x ?
    x_list = [[transform(x) for x in element[0:12]] for element in training_arr]
    # ??????
    test_data = [[transform(x) for x in element] for element in test_data]

    assert isinstance(y_list[0], int)  # y ??????? 0 ? 1, ? int
    assert len(x_list[0]) == 12  # x ????? 12 ???, ???? 12 ?????
    assert len(test_data[0]) == 12  # ????? x ??

    clf = GaussianNB()
    clf.partial_fit(x_list, y_list, numpy.unique(y_list))  # clf.fit(x_list, y_list) ?????
    res_arr = clf.predict(test_data)
    partial_fit_result = "".join([str(x) for x in res_arr])

    print("[*] ??????: {}".format(partial_fit_result))