#!/usr/bin/python3.7
# -*- coding: utf-8 -*-
# @Time : 2019/6/6 10:24
# @Author: Jtyoui@qq.com
"""统计指标：混淆矩阵
True Positive(真正，TP)：将正类预测为正类数
True Negative(真负，TN)：将负类预测为负类数
False Positive(假正，FP)：将负类预测为正类数
False Negative(假负，FN)：将正类预测为负类数
P=TP+FP 预测对的
N=FN+TN 预测错的
PP=TP+FN 实际上是对的
NN=FP+TN 实际上是错误的
ALL=P+N=PP+NN=所有
"""
[文档]def confusion_matrix(simple, pred):
"""混淆矩阵
:param simple: 实际样本分类列表
:param pred: 预测样本分类列表
:return: TP,TN,FP,FN
"""
tp, tn, fp, fn = 0, 0, 0, 0
for s, p in zip(simple, pred):
if s == p == 1: # 将正类预测为正类数
tp += 1
elif s == p == 0: # 将负类预测为负类数
tn += 1
elif s == 1 and p == 0: # 将正类预测为负类数
fn += 1
else: # 将负类预测为正类数
fp += 1
return tp, tn, fp, fn
[文档]def precision(tp, tn, fp, fn):
"""精确率=(TP+TN)/(TP+TN+FP+FN)"""
return (tp + tn) / (tp + tn + fp + fn)
[文档]def recall(tp, tn, fp, fn):
"""召回率=TP/(TP+FN)"""
return tp / (tp + fn)
[文档]def f_measure(tp, tn, fp, fn):
"""综合评价指标=(2*P*R)/(P+R)"""
p = precision(tp, tn, fp, fn)
r = recall(tp, tn, fp, fn)
return (2 * p * r) / (p + r)
if __name__ == '__main__':
simples = [1, 1, 0, 0, 0, 0, 1, 0, 0, 1] # 实际分类的值
prediction = [0, 1, 0, 1, 1, 0, 1, 1, 0, 1] # 预测分类的值
m = confusion_matrix(simple=simples, pred=prediction) # 混淆矩阵
p_ = precision(*m) # 准确率
r_ = recall(*m) # 召回率
f_ = f_measure(*m) # f值
print(p_, r_, f_) # 0.6 0.75 0.6666666666666665