This repository contains code and data for evaluating techniques in PDPilot.
- data: Downloading datasets, training models, computing PDP and ICE plots, and calculating feature importance scores.
- clustering: Evaluating the effect of preprocessing on the clustering results.
- ranking: Comparing feature importance rankings.
- filtering: Analyzing the effect of the tolerance parameter for filtering PDPs by shape.
- tool: Web app for labeling the shape of PDPs.
- environment.yml: conda environment file.
- local-requirements.txt: Exact versions of all packages that were used locally on my Mac.
- cluster-requirements.txt: Exact versions of all packages that were used on the Linux cluster.
conda env create -f environment.yml
conda activate pdpilot-eval