I trained data and generated a binary model file using xgboost R-package. Later I predicted from binary model file with my test data using both xgboost R-package and xgboost jvm-package, they matched!
When I predicted my test data using same binary model file with "xgboost-predictor-java", the predicted values didn't match with xgboost predicted results.
@komiya-atsushi I was doing mistake by setting treatsZeroAsNA parameter explicitly as false, for which I was getting wrong prediction values. Now the issue is resolved, Thanks!
I trained data and generated a binary model file using xgboost R-package. Later I predicted from binary model file with my test data using both xgboost R-package and xgboost jvm-package, they matched!
When I predicted my test data using same binary model file with "xgboost-predictor-java", the predicted values didn't match with xgboost predicted results.
Am I missing something here? Please find my test code here from pastebin: http://pastebin.com/Y6cUvRaX
Note: model: "gbtree" & function: "reg:linear"
cc: @komiya-atsushi