Native Bayes Find Ham Mail

import numpy
import random


def get_vac_list(data):
    vac_list = []
    for v in data:
        for vv in v:
            if len(vv) > 2:
                vac_list.append(vv)
    return list(set(vac_list))


def get_word_vec(sentence, vac_list):
    word_vec = [0]*len(vac_list)
    for v in sentence:
        if v in vac_list:
            word_vec[vac_list.index(v)] += 1
    return word_vec


def train(word_vec_list, class_list):
    n = len(word_vec_list)
    m = len(word_vec_list[0])
    p0vec = numpy.ones(m)
    p1vec = numpy.ones(m)
    p0sum = 2.0
    p1sum = 2.0
    for i in range(n):
        if class_list[i] == 0:
            p0vec += word_vec_list[i]
            p0sum += sum(word_vec_list[i])
        else:
            p1vec += word_vec_list[i]
            p1sum += sum(word_vec_list[i])
    p0class = numpy.log(p0vec/p0sum)
    p1class = numpy.log(p1vec/p1sum)
    pham = sum(class_list)/float(len(class_list))
    return p0class, p1class, pham


def classify(word_vec, p0class, p1class, pham):
    p0 = sum(p0class*word_vec)+numpy.log(1.0-pham)
    p1 = sum(p1class*word_vec)+numpy.log(pham)
    if p0 > p1:
        return 0
    else:
        return 1


def parse(str):
    import re
    data = re.split(r'\W*', str)
    data = [v.lower() for v in data if len(v) > 2]
    return data


def test():
    data = []
    class_list = []
    for i in range(1, 26):
        data.append(parse(open('email/ham/%d.txt' % i).read()))
        class_list.append(1)
        data.append(parse(open('email/spam/%d.txt' % i).read()))
        class_list.append(0)

    train_index_list = list(range(50))
    test_index_list = []
    for i in range(10):
        x = int(random.uniform(0, len(train_index_list)))
        test_index_list.append(train_index_list[x])
        del(train_index_list[x])

    train_data = []
    train_class_list = []
    for i in train_index_list:
        train_data.append(data[i])
        train_class_list.append(class_list[i])

    vac_list = get_vac_list(train_data)
    word_vec_list = []
    for v in train_data:
        word_vec_list.append(get_word_vec(v, vac_list))
    p0class, p1class, pham = train(word_vec_list, train_class_list)

    m = 0
    for i in test_index_list:
        if classify(get_word_vec(data[i], vac_list), p0class, p1class, pham) == class_list[i]:
            m += 1
    return m


if __name__ == '__main__':
    m = 0
    for i in range(1000):
        m += test()
    print(m)