Remove sanity/bounds check on predicted values

R/ensemble/h2oEnsemble-package/R/ensemble.R

@@ -224,8 +224,6 @@ predict.h2o.ensemble <- function(object, newdata) {
     }
   }
   names(basepreddf) <- names(object$basefits)
-  basepreddf[basepreddf < object$ylim[1]] <- object$ylim[1]  #Enforce bounds
-  basepreddf[basepreddf > object$ylim[2]] <- object$ylim[2]
   basepred <- as.h2o(localH2O, basepreddf, key="basepred")
 
   if (grepl("H2O", class(object$metafit))) {
@@ -235,9 +233,6 @@ predict.h2o.ensemble <- function(object, newdata) {
     # SuperLearner wrapper function metalearner
     pred <- predict(object=object$metafit$fit, newdata=basepred)
   }
-  # TO DO: Maybe restrict bounds here, but change code to work for pred as H2OParsedData obj
-  #pred[pred < object$ylim[1]] <- object$ylim[1]  #Enforce bounds
-  #pred[pred > object$ylim[2]] <- object$ylim[2]
   out <- list(pred = pred, basepred = basepred)
   return(out)
 }