Point of entry to the code

[joniali]

Hi, I was trying to run your code to recreate the results in your paper. I am not sure what is the point of entry to the code. I have tried the following things:

I ran all the following scripts with compas_violent_processed that you provided.

1. Run test_flipped_trianing, results in following error:

OSError: file data/compas_violent_processed.pickle not found

This expects the data to be in pickle format, while the data in the repository is in csv format. Do I need to save it in pickle?

2. Running exp ex_solve_disc_mip results in following error:

File "scripts/ex_solve_disc_mip.py", line 35, in
assert baseline_file_name.exists()
AssertionError

3. Running test_mip, pointing to the dataset you have uploaded, results in following error:

File "eqm/tests/test_mip.py", line 36, in
mip = ZeroOneLossMIP(data = data, settings = test_settings)
File "/predictive_multiplicity/eqm/mip.py", line 527, in init
cpx, info = build_mip(self.data, settings = self.SETTINGS, var_name_fmt = self.VAR_NAME_FMT, **kwargs)
TypeError: build_mip() got multiple values for keyword argument 'settings'

I am using Python 3.7 and I had to correct the imports, i.e., I had to add the eqm directory in the path variable.

Should the code run out of the box with the given data or do I have to change something? Any help you can provide would be great!

[gagb]

Hey Berk/Charlie, I would really appreciate more structure in the README and instructions on how to run the code!
And, thanks for releasing the code BTW! :)

-G

[charliemarx]

I added a short example script in "example.py". Let me know if you have any trouble running that code or have any other questions and I'd be happy to help.