如何查看递归特征消除（RFE）生成的每个模型的预测变量？

发布于 2025-01-17 02:12:05 字数 529 浏览 4 评论 0原文

我使用 RFE（递归特征消除）方法从 39 个变量中找到最佳模型。我使用了以下代码：

set.seed(10)
ctrl <- rfeControl(functions = caretFuncs, method = "repeatedcv",repeats = 5,number= 5,allowParallel = TRUE)

RF_39 <- rfe(X, Y,sizes = c(1:39),method ='rf',rfeControl = ctrl,tuneGrid = data.frame(mtry=6))

最好的模型是使用 36 个变量构建的。如果我想查看这 36 个变量，我可以使用 预测器(RF_39) 或 RF_39$optVariable 函数。

但是，我需要查看用于制作其他模型而不是最佳模型的变量。例如，用于制造型号 12 的变量是什么？如何查看RFE方法制作的特定模型的变量？感谢您的帮助。

原文

I have used RFE (Recursive Feature Elimination) method to find the best model out of 39 variables. I used the following codes:

set.seed(10)
ctrl <- rfeControl(functions = caretFuncs, method = "repeatedcv",repeats = 5,number= 5,allowParallel = TRUE)

RF_39 <- rfe(X, Y,sizes = c(1:39),method ='rf',rfeControl = ctrl,tuneGrid = data.frame(mtry=6))

The best model is made using 36 variables.
If I want to see those 36 variables I can use
predictors(RF_39) or RF_39$optVariable functions.

However, I need to see the variables used for making the other models rather than the best model.
For example, what are the variables used for making model number 12?
How can I see the variables of a specific model made by RFE method?
Thanks for your help.

分享到QQ

分享到微博