INT104-lab7

Python 信息熵-条件熵计算

import csv
import numpy as np


def read(path: str) -> tuple:
    with open(path, 'r') as f:
        text = csv.reader(f)
        A = []
        for row in text:
            A.append(row)
    return A, len(A), len(A[0])


def h(p: float) -> float:
    return -p * np.log2(p)


def Hc(A: list, N: int, M: int) -> float:
    count = 0
    for a in A:
        count += int(a[M - 1])
    p = count / N
    return h(p) + h(1 - p)


def Hcw(A: list, w: int, N: int, M: int) -> float:
    subset0 = []
    subset1 = []
    for a in A:
        if int(a[w]) == 0:
            subset0.append(a)
        else:
            subset1.append(a)
    p0 = (len(subset0) / N)
    p1 = (len(subset1) / N)
    return p0 * Hc(subset0, N, M) + p1 * Hc(subset1, N, M)


if __name__ == '__main__':
    path = 'binary_data.csv'
    A, N, M = read(path)
    for i in range(5):
        IG = Hc(A, N, M) - Hcw(A, i, N, M)
        print("IG(w = %d) = %.5f" % (i, IG))