深度学习: 回归分析 评价方法

MAE

1n∑ni=1∣y−y^∣1n∑i=1n∣y−y^∣\frac{1}{n}\sum_{i=1}^n\mid{y}-\hat{y}\mid

MSE

1n∑ni=1⟮y−y^⟯21n∑i=1n⟮y−y^⟯2\frac{1}{n}\sum_{i=1}^n{\lgroup{y}-\hat{y}\rgroup}^2

MSE==RMSE2MSE==RMSE2{MSE} == {RMSE}^2

RMSE

1n∑ni=1⟮y−y^⟯2−−−−−−−−−−−−−√1n∑i=1n⟮y−y^⟯2\sqrt{\frac{1}{n}\sum_{i=1}^n{\lgroup{y}-\hat{y}\rgroup}^2}

RMSE==MSE−−−−−√RMSE==MSE{RMSE} == \sqrt{MSE}