Model Evaluation tab issue

Issue #28 new
Paul Levchuk created an issue

It seems that I found bug. When I doing Model Evaluation I got sometimes error:

Error in rattle::errorMatrix(crs$dataset[crs$validate, c(crs$input, crs$target)]$BK2_lead_conversion, : The supplied actual and predicted must have the same levels.

I copied code from Log to RStudio and fixed it specifying levels from dataset:

Obtain the response from the Extreme Boost model.

lvls <- levels(as.factor(crs$dataset[[crs$target]])) crs$pr <- factor(ifelse(predict(crs$ada, crs$dataset[crs$test, c(crs$input, crs$target)]) > 0.5, lvls[2], lvls[1]), levels = levels(crs$dataset[crs$test, c(crs$input, crs$target)]$BK2_lead_conversion))

Is it possible to add this fix to rattle?

Comments (0)

  1. Log in to comment