From the "Intro to Caret" vignette (pdf) you can download from the caret CRAN page, I am tying to run a simple example with built in data.
I get the following error:
Error in { :
task 1 failed - "unused arguments (type = "prob", ncomp = modelFit$tuneValue$ncomp)"
In addition: There were 30 warnings (use warnings() to see them)
And all the warnings() seem to follow this pattern:
30: In eval(expr, envir, enclos) :
predictions failed for Fold10.Rep3: ncomp=15 Error in predict(modelFit, newdata, type = "class") :
unused argument (type = "class")
I have other caret examples working, this seems to come up with logistic regression problems that are trying to predict a factor. You should be able to reproduce with:
library(mlbench)
library(caret)
data(Sonar)
set.seed(1)
inTrain <- createDataPartition(y = Sonar$Class, p = 0.75, list = FALSE)
training <- Sonar[inTrain,]
testing <- Sonar[-inTrain,]
ctrl <- trainControl(method = "repeatedcv",
repeats = 3,
classProbs = TRUE,
summaryFunction = twoClassSummary)
model <- train(Class ~ .,
data = training,
method = "pls",
tuneLength = 15,
trControl = ctrl,
metric = "ROC",
preProc = c('center', 'scale'))
I am using R Version 3.3
Any ideas interpreting this error message? Any other posts seem to point to format of the data.frame, but this is built in data and a documented example.
Based on feedback in comments that it was working on other platforms, I chose to wipe out R and reinstall and reinstall all packages, and this resolved. I do occasionally see it come up and it seems to help to clear the workspace.