CODEWITHSUNDEEP
rr-caret
Stanley
Stanley

Reputation: 2806

R caret regression

Trying to use R caret to train a model using lm linear regression:

Code is as per below:

modelLm <- train(x=X_train,y=Y_train, method="lm", na.action = na.omit, trControl=control)

But get the error: 

Error in quantile.default(y, probs = seq(0, 1, length = cuts)) : 
            missing values and NaN's not allowed if 'na.rm' is FALSE

Isn't na.action = na.omit supposed to ignore the missing values?

Upvotes: 0

Views: 1192

Answers (1)

topepo
topepo

Reputation: 14316

When you don't use a formula, the train calls train.default and na.action is not one of its arguments. na.action is nicely baked into the formula machinery so either use that interface or use complete.cases to get rid of them for train.default.

Max

Upvotes: 1

Related Questions