[Python知识库]ID3算法 信息熵计算公式

数据挖掘课程作业题，信息熵还用手算，我心想这不是累傻小子吗？

于是用python读表，让电脑自己算去。

应该根据算出来的结果选取最大值继续算的，懒得写了，先发上来，哪天心情好补上。

gain.py

import math

def I(s1, s2):
    '''
    :param s1: 值为1的数量
    :param s2: 值为0的数量
    :return: 返回期望值
    '''
    s = s1 + s2
    if s1 == 0 or s2 == 0:
        return 0
    # print("s1 = {}, s2 = {}, s = {}".format(s1,s2,s))
    ex = - (s1 / s) * math.log(s1 / s, 2) - (s2 / s) * math.log(s2 / s, 2)
    print("I = {}".format(ex))
    return ex

def E(s1, s2, s11, s21, s12, s22):
    '''
    :param s1: 属性1的数量
    :param s2: 属性0的数量
    :param s11: 需要计算的属性1中最终分类为1的数量
    :param s21: 需要计算的属性1中最终分类为0的数量
    :param s12: 需要计算的属性0中最终分类为1的数量
    :param s22: 需要计算的属性0中最终分类为0的数量
    :return: 返回计算出的根据属性A划分出的熵值
    '''
    s = s1 + s2
    if s11+s12+s21+s22 != s:
        print("Error param! Please check!")
        exit(-1)
    entropy = (s1 / s) * I(s11, s21) + (s2 / s) * I(s12, s22)
    print("E(A) = {}".format(entropy))
    return entropy

def Gain(s):
    '''
    :param s: 列表，含8个参数：
                                0 [长毛鸡数量] 1 [不长毛鸡数量]
                                2 [长毛鸡下蛋数量] 3 [长毛鸡不下蛋数量]
                                4 [不长毛鸡下蛋数] 5 [不长毛鸡不下蛋数]
                                6 [下蛋鸡总数] 7 [不下蛋鸡总数]
    :return:
    '''
    if len(s) != 8:
        print("The third param must be a list as 8 member")
        exit(-1)
    gain = I(s[6], s[7]) - E(s[0], s[1], s[2], s[3], s[4], s[5])
    print("Gain = {}\n".format(gain))
    return gain

主函数

main.py

import numpy as np
from gain import Gain


def read_csv():
    p = r'./layegg.csv'
    with open(p, encoding='utf-8') as f:
        data = np.loadtxt(f, int, delimiter=",", skiprows=1)
    return data


def find_col(tb):
    s6,s7 = 0,0
    for i in range(tb.shape[0]):
        if tb[i][tb.shape[1] - 1] == 1:
            s6 = s6 + 1     # 下蛋的鸡
        else:
            s7 = s7 + 1     # 不下蛋的鸡
    gain_list = []

    for j in range(1, tb.shape[1] - 1):
        s11, s21, s12, s22, s_1, s_2 = 0, 0, 0, 0, 0, 0
        for i in range(tb.shape[0]):
            if tb[i][j] == 1:  # 长毛
                if tb[i][tb.shape[1] - 1] == 1:  # 长毛鸡下蛋
                    s11 = s11 + 1
                else:  # 长毛鸡不下蛋
                    s21 = s21 + 1
                s_1 = s_1 + 1  # 长毛鸡的数量
            else:  # 不长毛
                if tb[i][tb.shape[1] - 1] == 1:  # 不长毛鸡下蛋
                    s12 = s12 + 1
                else:
                    s22 = s22 + 1
                s_2 = s_2 + 1  # 不长毛鸡不下蛋
        s_list = [s_1, s_2, s11, s21, s12, s22, s6, s7]
        print(s_list)
        gain_list.append([j, Gain(s_list)])
    return gain_list


if __name__ == '__main__':
    tb = read_csv()
    gain_list = find_col(tb)
    print(gain_list)

layegg.csv