python实现混淆矩阵（numpy）

假设有A、B、C、D、E五个类别

一 确定输出和标签

import numpy as np
y_true = ["A"]*20 + ["B"]*20 + ["C"]*20 + ["D"]*20 + ["E"]*20

len(y_true)

y_true

y_pred =["D"]*5+ ["A"]*30 + ["B"]*15 + ["C"]*25 + ["D"]*20 + ["E"]*5

len(y_pred)

y_pred

二 初始化混淆矩阵

# 初始化混淆矩阵
classes_num = 5
confusion_matrix = np.zeros((classes_num,classes_num))

三 计算混淆矩阵思路：同时遍历所有pred和label,发现属于混淆矩阵哪一个位置，哪一个位置就+1

step1:将pred和label进行一对一组合

list_pred_label = []
for i in range(len(y_pred)):
  list_pred_label.append([y_pred[i],y_true[i]])

print(list_pred_label)

Step2:遍历list_pred_label，将其中的类别转为混淆矩阵索引（A:0,B:1,C:2,D:3,E:4）

for i in list_pred_label:
  if i[0] == "A":
    i[0] = 0
  elif i[0] == "B":
    i[0] = 1
  elif i[0] == "C":
    i[0] = 2
  elif i[0] == "D":
    i[0] = 3  
  elif i[0] == "E":
    i[0] = 4  
  if i[1] == "A":
    i[1] = 0
  elif i[1] == "B":
    i[1] = 1
  elif i[1] == "C":
    i[1] = 2
  elif i[1] == "D":
    i[1] = 3  
  elif i[1] == "E":
    i[1] = 4

print(list_pred_label)

step3:对混淆矩阵进行赋值

for i in list_pred_label:
  confusion_matrix[i[0],i[1]] += 1

print(confusion_matrix)