Caret训练函数中的PCA预处理参数。

Question

我正在进行 knn 在我的数据上进行回归，并希望。

a) 通过交叉验证 repeatedcv 来寻找一个最佳 k;

b) 在建立knn模型时，使用 PCA 在 90% 级阈值来降低维度。

library(caret)
library(dplyr)
set.seed(0)
data = cbind(rnorm(20, 100, 10), matrix(rnorm(400, 10, 5), ncol = 20)) %>% 
  data.frame()
colnames(data) = c('True', paste0('Day',1:20))
tr = data[1:15, ] #training set
tt = data[16:20,] #test set

train.control = trainControl(method = "repeatedcv", number = 5, repeats=3)
k = train(True ~ .,
          method     = "knn",
          tuneGrid   = expand.grid(k = 1:10), 
          #trying to find the optimal k from 1:10
          trControl  = train.control, 
          preProcess = c('scale','pca'),
          metric     = "RMSE",
          data       = tr)

我的问题是

(1) 我注意到 某人 建议修改trainControl中的pca参数。

ctrl <- trainControl(preProcOptions = list(thresh = 0.8))
mod <- train(Class ~ ., data = Sonar, method = "pls",
              trControl = ctrl)

如果我修改了trainControl中的参数，是否意味着在KNN过程中仍然进行PCA？和这个问题类似

(2)我发现了另一个 例子 这符合我的情况--我希望将阈值改为90%，但我不知道在哪里可以将它改成这样。Caret's train 功能，尤其是我还需要 scale 选项。

我为我冗长的描述和随机的参考资料道歉。先谢谢你!

(感谢Camille的建议，使代码能正常工作！)

Answer 1

回答你的问题。

我注意到有人建议改变trainControl中的pca参数。

mod <- train(Class ~ ., data = Sonar, method = "pls",trControl = ctrl)

如果我改变了trainControl中的参数，是否意味着在KNN过程中仍然要进行PCA？

是的，如果你用它来做。

train.control = trainControl(method = "repeatedcv", number = 5, repeats=3,preProcOptions = list(thresh = 0.9))

k = train(True ~ .,
          method     = "knn",
          tuneGrid   = expand.grid(k = 1:10), 
          trControl  = train.control, 
          preProcess = c('scale','pca'),
          metric     = "RMSE",
          data       = tr)

你可以在preProcess下检查。

k$preProcess
Created from 15 samples and 20 variables

Pre-processing:
  - centered (20)
  - ignored (0)
  - principal component signal extraction (20)
  - scaled (20)

PCA needed 9 components to capture 90 percent of the variance

这样就可以回答2)，就是单独使用preProcess。

mdl = preProcess(tr[,-1],method=c("scale","pca"),thresh=0.9)
mdl
Created from 15 samples and 20 variables

Pre-processing:
  - centered (20)
  - ignored (0)
  - principal component signal extraction (20)
  - scaled (20)

PCA needed 9 components to capture 90 percent of the variance

train.control = trainControl(method = "repeatedcv", number = 5, repeats=3)

k = train(True ~ .,
          method     = "knn",
          tuneGrid   = expand.grid(k = 1:10), 
          trControl  = train.control,
          metric     = "RMSE",
          data       = predict(mdl,tr))