25、使用交叉验证调参cross_val_score

东方耀

25、使用交叉验证调参cross_val_score


训练数据用于模型训练
验证数据用于模型评判并调参
测试数据集不参与模型的创建与优化，对模型来说是完全新的数据，是未知情况




ipynb文件在附件，可提供下载！

1. # 使用交叉验证来调参
   
2. from sklearn.model_selection import cross_val_score
   
3. knn_clf = KNeighborsClassifier()
   
4. cross_val_score(knn_clf, X_train, y_train, cv=5)
   
5. %%time
   
6. best_score, best_k, best_p = 0, 0, 0
   
7. for k in range(2, 11):
   
8.     for p in range(1, 6):
   
9.         knn_clf = KNeighborsClassifier(weights='distance', n_neighbors=k, p=p)
   
10.         scores = cross_val_score(knn_clf, X_train, y_train)
    
11.         score = np.mean(scores)
    
12.         if score > best_score:
    
13.             best_score = score
    
14.             best_k = k
    
15.             best_p = p
    
16. print(best_score, best_k, best_p)
    
17. 
    
18. best_knn_clf = KNeighborsClassifier(weights='distance', n_neighbors=2, p=2)
    
19. best_knn_clf.fit(X_train, y_train)
    
20. best_knn_clf.score(X_test, y_test)
    
21. 
    
22. best_knn_clf = grid_search.best_estimator_
    
23. best_knn_clf.score(X_test, y_test)

复制代码

视频教程请参考：http://www.ai111.vip/thread-349-1-1.html
东方老师微信：dfy_88888

fglbee

this is good idea