本文主要是介绍ID3算法 信息熵计算公式，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！
gain.py
import mathdef 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 + s2if 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 exdef 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 + s2if 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 entropydef 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 Gaindef read_csv():p = r'./layegg.csv'with open(p, encoding='utf-8') as f:data = np.loadtxt(f, int, delimiter=",", skiprows=1)return datadef find_col(tb):s6,s7 = 0,0for 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, 0for 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 + 1else: # 长毛鸡不下蛋s21 = s21 + 1s_1 = s_1 + 1 # 长毛鸡的数量else:这篇关于ID3算法 信息熵计算公式的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！
