python分类算法评估模型_机器学习中分类和回归模型的评价指标

分类算法的效果评估

1，准确率accuracy_score

from sklearn.metrics import accuracy_score

2，精确率/查准率precision_score

from sklearn.metrics import precision_score

分为宏平均(macro)和微平均(micro)，宏平均比微平均更合理。

metrics.precision_score(y_true, y_pred, average='micro')

metrics.precision_score(y_true, y_pred, average='macro')

其中average参数有五种：(None, ‘micro’, ‘macro’, ‘weighted’, ‘samples’)

3，召回率/查全率recall_score

from sklearn.metrics import recall_score

召回率也有宏平均和微平均的区别，和上面的用法一样。

4，F1-score

from sklearn.metrics import f1_score

metrics.f1_score(y_true, y_pred, average='weighted')

5，混淆矩阵(confusion-matrix)

from sklearn.metrics import confusion_matrix

6，分类报告(classification_report)

from sklearn.metrics import classification_report

包含precision/recall/f1-score/均值/分类个数

7，kappa score

from sklearn.metrics import cohen_kappa_score

cohen_kappa_score(y_true, y_pred)

8，ROC

1)计算ROC值

from sklearn.metrics import roc_auc_score

roc_auc_score(y_true, y_scores)

2)画ROC图

具体画ROC图的方法请参照官方给出的代码

http://scikit-learn.org/stable/auto_examples/model_selection/plot_roc.html

9，距离

1)海明距离(hamming_loss)

from sklearn.metrics import hamming_loss

hamming_loss(y_true, y_pred)

2)Jaccard距离(jaccard_similarity_score)

from sklearn.metrics import jaccard_similarity_score

jaccard_similarity_score(y_true, y_pred)

回归算法的评价指标

1，可释方差也叫解释方差(explained_variance_score)

from sklearn.metrics import explained_variance_score

explained_variance_score(y_true, y_pred)

2，平均绝对误差(mean_absolute_error)

from sklearn.metrics import mean_absolute_error

mean_absolute_error(y_true, y_pred)

3，均方误差(mean_squared_error)

from sklearn.metrics import mean_squared_error

mean_squared_error(y_true, y_pred)

4，中值绝对误差(median_absolute_error)

from sklearn.metrics import median_absolute_error

median_absolute_error(y_true, y_pred)

5，R方值，确定系数(r2_score)

from sklearn.metrics import r2_score

r2_score(y_true, y_pred)

作者：曦宝

链接：https://www.jianshu.com/p/c3cf5c6081ad