XGBoost Error

Question

Hi All ,

I downloaded the XGBoost macro and ditched the python section to stick with R. I can run the workflow with no issue. If I modify the inputs to then represent my own data set, I keep getting E"rror: XGBoost_R (517): Error in if ((maximize && score > best_score) ||Tool #1: (!maximize && score < :" coming back. Any ideas (using regression, not classification)

I've ensured I have no text fields, nulls, errors, blanks etc. Everything going in is set to double.

What is this R error telling me?

Thanks

mceleavey · Answer

Hi @stevehughes ,

Yeah, without seeing the data it's going to be difficult.

You're trying to maximise a specific field, which I assume is your target variable. Is this a numeric field, containing no non-numeric fields, and you have enough rows in your data to support the calculation?

M.

stevehughes · Answer

Apologies for my tardy reply. I basically took the downloaded macro and and swapping out the training and scoring data. I've forced all columns to either INT or DBL as per the working download. No nulls or text strings anywhere in the data. Only alteration I made to the XGB config was to choose the regression option as I'm trying to predict sales.

Currently, my scoring set is the same as the training set (for testing) just with the sales column removed of course.

Constantly getting:

Error: XGBoost_R (517): Error in if ((maximize && score > best_score) ||Tool #1: (!maximize && score < :

..and my R knowledge is shabby at best!

ax_config.png

ax_inputs.png

ax_error.png

mceleavey · Answer

Hi @stevehughes ,

without seeing the data going into the model and the config of the model it's going to be very difficult for us to help.

M.