R：在 Logistic 回归上使用 Caret 进行交叉验证的特征选择

Question

我目前正在学习如何在 R 中实现逻辑回归

我已经获取了一个数据集并将其拆分为训练和测试集，并希望使用交叉验证来实现 forward selection、backward selection 和 best subset selection 以选择最佳特征。 我正在使用caret 在训练数据集上实现cross-validation，然后在测试数据上测试预测。

我在插入符号中看到了rfe 控件，还查看了caret website 上的文档以及问题How to use wrapper feature selection with algorithms in R? 上的链接。我不清楚如何更改特征选择的类型，因为它似乎默认为向后选择。任何人都可以帮助我完成我的工作流程。下面是一个可重现的例子

library("caret")

# Create an Example Dataset from German Credit Card Dataset
mydf <- GermanCredit

# Create Train and Test Sets 80/20 split
trainIndex <- createDataPartition(mydf$Class, p = .8, 
                              list = FALSE, 
                              times = 1)

train <- mydf[ trainIndex,]
test  <- mydf[-trainIndex,]


ctrl <- trainControl(method = "repeatedcv", 
                 number = 10, 
                 savePredictions = TRUE)

mod_fit <- train(Class~., data=train, 
             method="glm", 
             family="binomial",
             trControl = ctrl, 
             tuneLength = 5)


# Check out Variable Importance
varImp(mod_fit)
summary(mod_fit)

# Test the new model on new and unseen Data for reproducibility
pred = predict(mod_fit, newdata=test)
accuracy <- table(pred, test$Class)
sum(diag(accuracy))/sum(accuracy)

Answer 1

你可以简单地在 mod_fit 中调用它。当涉及到逐步后退时，下面的代码就足够了

trControl <- trainControl(method="cv",
                          number = 5,
                          savePredictions = T,
                          classProbs = T,
                          summaryFunction = twoClassSummary)

caret_model <- train(Class~.,
                     train,
                     method="glmStepAIC", # This method fits best model stepwise.
                     family="binomial",
                     direction="backward", # Direction
                     trControl=trControl)

注意在trControl中

method= "cv", # No need to call repeated here, the number defined afterward defines the k-fold.
classProbs = T,
summaryFunction = twoClassSummary # Gives back ROC, sensitivity and specifity of the chosen model.