How to load dataset (orc, parquet) to castable and get deterministic results when training

[claudiodejesusa]

Hi everyone,

I'm working with data-files in format .orc or .parquet, but when it is loaded to castable and then train a model (proc gradboost for example), i got different results every time i run the code. This situation, does not happen when i try with format .sashdat and train several times the same model.

I need suggestion about it, how to get deterministic results with format .orc or .parquet because the storage - by default - my job decide is ADLS2.

Thanks,

Claudio