R語言模型選擇之精度准則與最大值法問題

在模型選擇中我們一般用caret包train函數建立模型，並對模型進行評判

方法1：

set.seed(1234)
tr_control<-trainControl(method = 'cv',number = 5)
# 創建隨機森林模型
model_rf<-train(Class~.,data=traindata,
                trControl=tr_control,method='rf')
model_rf

輸出

mtry Accuracy Kappa
2 0.9276465 0.8552977
16 0.9314521 0.8628921
30 0.9276627 0.8553120

Accuracy was used to select the optimal model using the largest value.
The final value used for the model was mtry = 16.

方法2

set.seed(1234)
model_rf <- train(Class ~., data = traindata, method = 'rf', 
                  trControl = trainControl(method = 'cv', 
                                           number = 5, 
                                           selectionFunction = 'oneSE'))
model_rf

mtry Accuracy Kappa
2 0.9276143 0.8552365
16 0.9212771 0.8425685
30 0.9250988 0.8502003

Accuracy was used to select the optimal model using the one SE rule.
The final value used for the model was mtry = 2.

 

可以看到二者選定的模型並不一樣，而且選定的標准也不一樣，方法1標准是最大值法，方法2是精確度。

原因在方法2中用了：selectionFunction = 'oneSE'