Make a custom trainControl
The wine quality dataset was a regression problem, but now you are looking at a classification problem. This is a simulated dataset based on the "don't overfit" competition on Kaggle a number of years ago.
Classification problems are a little more complicated than regression problems because you have to provide a custom summaryFunction
to the train()
function to use the AUC
metric to rank your models. Start by making a custom trainControl
, as you did in the previous chapter. Be sure to set classProbs = TRUE
, otherwise the twoClassSummary
for summaryFunction
will break.
This exercise is part of the course
Machine Learning with caret in R
Exercise instructions
Make a custom trainControl
called myControl
for classification using the trainControl
function.
- Use 10 CV folds.
- Use
twoClassSummary
for thesummaryFunction
. - Be sure to set
classProbs = TRUE
.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Create custom trainControl: myControl
myControl <- trainControl(
method = "cv",
number = ___,
summaryFunction = ___,
classProbs = ___, # IMPORTANT!
verboseIter = TRUE
)