模型结果验证方式

最新推荐文章于 2024-05-29 11:40:23 发布

E.K.江湖念书人

最新推荐文章于 2024-05-29 11:40:23 发布

阅读量111

点赞数

文章标签：数学建模

本文链接：https://blog.csdn.net/weixin_53704902/article/details/131104673

版权

import datasets  
df = df_test
ref = []
pred = []
for i in df.index:
  if str(df.loc[i,'predict'])!= 'nan':
    ref.append(df.loc[i,'feature_label']) 
    pred.append(df.loc[i,'predict']) 
acc_metric = datasets.load_metric("accuracy")
f_metric = datasets.load_metric("f1")
r_metric = datasets.load_metric('recall')
p_metric = datasets.load_metric('precision')

result = dict()
result.update(acc_metric.compute(predictions=pred, references=ref))
result.update(f_metric.compute(predictions=pred, references=ref,average="macro"))
result.update(p_metric.compute(predictions=pred, references=ref,average="macro"))
result.update(r_metric.compute(predictions=pred, references=ref,average="macro"))
df_test.to_excel('example.xlsx')
print(result)