模型评估

 1 # -*- coding: utf-8 -*-
 2 """
 3 Created on Thu Sep 27 16:24:29 2018
 4     模型及预测准确度评估
 5 @author: zhen
 6 """
 7 
 8 from sklearn import metrics
 9 
10 if __name__ == "__main__":
11     
12     # 同一性homogeneity：每个群集只包含单个类的成员。 
13     # 完整性completeness：给定类的所有成员都分配给同一个群集。
14     # 调和平均V-measure
15     y = [0, 0, 0, 1, 1, 1]
16     y_hat = [0, 0, 1, 1, 2, 2]
17     h = metrics.homogeneity_score(y, y_hat)
18     c = metrics.completeness_score(y, y_hat)
19         
20     v2 = 2 * c * h / (c + h)
21     v = metrics.v_measure_score(y, y_hat)
22     print(u‘同一性(Homogeneity):‘, h)
23     print(u‘完整性(Completeness):‘, c)
24     print(u‘V_Measure:‘, v2, v)
25     
26     y = [0, 0, 0, 1, 1, 1]
27     y_hat = [0, 0, 1, 3, 3, 3]
28     h = metrics.homogeneity_score(y, y_hat)
29     c = metrics.completeness_score(y, y_hat)
30     v = metrics.v_measure_score(y, y_hat)
31     
32     print(u‘同一性(Homogeneity):‘, h)
33     print(u‘完整性(Completeness):‘, c)
34     print(u‘V_Measure:‘, v)
35     
36     # 允许不同值
37     y = [0, 0, 0, 1, 1, 1]
38     y_hat = [1, 1, 1, 0, 0, 0]
39     h = metrics.homogeneity_score(y, y_hat)
40     c = metrics.completeness_score(y, y_hat)
41     v = metrics.v_measure_score(y, y_hat)
42     
43     print(u‘同一性(Homogeneity):‘, h)
44     print(u‘完整性(Completeness):‘, c)
45     print(u‘V_Measure:‘, v)
46     
47     # 兰德指数
48     # ARI值的范围是[-1,1]，负的结果都是较差的，越接近-1表示聚类越差，
49     # 正的结果都是较好的，1是最佳结果，越接近1表示聚类结果越好。
50     y = [0, 0, 1, 1]
51     y_hat = [0, 1, 0, 1]
52     ari = metrics.adjusted_rand_score(y, y_hat)
53     print("兰德指数:", ari)
54     
55     y = [0, 0, 0, 1, 1, 1]
56     y_hat = [0, 0, 1, 1, 2, 2]
57     ari = metrics.adjusted_rand_score(y, y_hat)
58     print("兰德指数:", ari)
59