====== 评价指标 ======
ML benchmarks [[https://github.com/szilard/benchm-ml]]\\
===F1-score===
===Youden Index===
===AIC=== 
===BIC===
=====回归评价=====
[[https://blog.csdn.net/skullFang/article/details/79107127]]\\
===均方误差（MSE）===
用 真实值-预测值 然后平方之后求和平均。\\
<code>
y_preditc=reg.predict(x_test) #reg是训练好的模型
mse_test=np.sum((y_preditc-y_test)**2)/len(y_test) #跟数学公式一样的
</code>
===均方根误差（RMSE）===
RMSE（Root Mean Squard Error）均方根误差。\\
<code>
rmse_test=mse_test ** 0.5
</code>

===MAE(平均绝对误差) ===
<code>
mae_test=np.sum(np.absolute(y_preditc-y_test))/len(y_test)
</code>
===R Squared===
分子是Residual Sum of Squares 分母是 Total Sum of Squares 
分子就变成了我们的均方误差MSE，下面分母就变成了方差。
<code>
1- mean_squared_error(y_test,y_preditc)/ np.var(y_test)
</code>
===sklearn===
<code>
from sklearn.metrics import mean_squared_error #均方误差
from sklearn.metrics import mean_absolute_error #平方绝对误差
from sklearn.metrics import r2_score#R square
#调用
mean_squared_error(y_test,y_predict)
mean_absolute_error(y_test,y_predict)
r2_score(y_test,y_predict)
</code>
====不平衡数据====
[[https://www.kaggle.com/lct14558/imbalanced-data-why-you-should-not-use-roc-curve/notebook]]\\