Installation (Cross-Platform)

First install Poetry. Open a terminal, cd to the cloned repository directory, and run

poetry install

to setup and install the virtual environment.

Basic Testing

WARNING: Tests will use all your available cores and spew a huge amount of text to your terminal, so only run them if you are prepared.

poetry run pytest

Running the Paper Analyses

Compute Canada

The code to run the analyses here is unfortunately deeply tied up with Compute Canada cluster specifics and SLURM. Normally, once you have activated your virtual environment with e.g. poetry shell, the procedure would be to run:

python analysis/create_jobscripts.py
sbatch analysis/job_scripts/submit_all_downsampling.sh
sbatch analysis/job_scripts/submit_all_feature.sh
sbatch analysis/job_scripts/submit_mlp_downsampling.sh
sbatch analysis/job_scripts/submit_mlp_feature.sh

There are hard-coded switches that modify resource requests and runtimes depending on the cluster, so reproducing this would unfortunately require reading and modifying the code.

Running Locally

Alternately, you can run a specific dataset analysis by specifying command-line arguments and faking the environment. For example (assuming you have run poetry shell):

CC_CLUSTER=niagara \
python analysis/feature_downsampling.py \
  --classifier=lr \
  --dataset=diabetes \
  --kfold-reps=50 \
  --n-percents=200 \
  --results-dir=<your_directory_here> \
  --cpus=8 \
  --pbar

Don't expect this to work on Windows, and it is most likely unfeasible to run all analyses like this on a single machine.

Hypertuning can likely be done locally without too much trouble. For example, to tune the logistic regression classifier, run:

poetry run pytest tests/test_analysis_hypertune.py::test_lr_params