My attempts for the above specified challenge. To run the code, do create the following sub-directories santander
- data
- train.csv
- test.csv
- plots
- models
- ref
- Spanish2English.csv (for translating spanish column names to English)
In addition, you might want to ensure that you have the relevant dependencies/packages installed.
You can then run the main script which does the cross-validation for parameter searching with the following:
Rscript xgb_cv.R
To blend the final submissions together, run the following on the command line:
Rscript ensembler.R
Too bad I was a little lazy and didn't ensemble my models for the final submission. It would have gotten me the 18th place on the private LB:
