CODEWITHSUNDEEP
rr-caretglmnet
ZJAY
ZJAY

Reputation: 2837

Tuning for alpha/lambda using Caret

My end goal, with data in the matrix below, is training a number of models across a grid of different lambdas and alphas, using the glmnet method. Perhaps there is another way to approach this tuning problem as well.

x <- Macro[1:13, 3:21]
x <- as.matrix(x)
y <- Macro[1:13, 2:2]
y <- as.matrix(y)

myfit <- caret::train(x, y,
                  method = "glmnet",
                  tuneGrid = expand.grid(.alpha = seq(.05, 1, length = 15),
                                                     .lambda = c((5:10)/10)))

The above code returns the following error:

Error in train.default(x, y, method = "glmnet", tuneGrid = expand.grid(.alpha = seq(0.05, : Metric RMSE not applicable for classification models

enter image description here

Upvotes: 2

Views: 1010

Answers (1)

Pol Ferrando
Pol Ferrando

Reputation: 673

The problem is that your response variable y is a matrix but it should be "A numeric or factor vector containing the outcome for each sample" according to the documentation of train. Therefore, you just have to remove y <- as.matrix(y) from your code and it will work.

Upvotes: 1

Related Questions