Run Benchmark for custom model

[AnFreTh]

Hi,

first of all, great Code/Paper!
I wanted to run a simple model to compare it with your results. Is there an easy way to for example run a simple sklearn model such that I can directly compare to the results reported in your paper?

[puhsu]

Hi! I'm glad to hear that you are interested in the benchmark!

We plan to add a more streamlined way to setup and process the datasets.

Until then the steps to use TabReD are as follows:

1. Create an env (following readme instruction)
2. Run mkdir data
3. Run python preprocessing/<dataset-name>.py for each dataset (it should be quick, longest parts are the downloads)
4. Modify any of the bin scripts appropriate. I suggest looking at the bin/xgboost.py and switching the model for the one you would like to test (as the GBDT implementations are sklearn-api compatible).

I'll keep the issue open for now, until we make a more streamlined setup for dataset preparation