FeatureEngineering bug

[jreps]

PatientLevelPrediction/R/CyclopsModels.R

Line 191 in c5176a2

featureEngineering = attr(trainData, "metaData")$featureEngineering,#learned mapping

should be extracting from trainData$covariateData.

This is repeated in all the model fitting functions and needs to be updated.

[egillax]

Just noting this here for myself, I was confused by the bug above so I broke the example vignette for feature engineering (only on develop) few weeks ago. That needs fixing as well: https://github.com/OHDSI/PatientLevelPrediction/blob/develop/vignettes/AddingCustomFeatureEngineering.Rmd#L146

And in the new imputation methods I'm adding in branch imputation I need to remember to be consistent with attaching the featureEngineering stuff to trainData$covariateData . Would be good to add tests to enforce this consistency as well.