Software and data underlying the publication: Endogenous Macrodynamics in Algorithmic Recourse

Code and research results for SaTML 2023 research paper. Originally released here: https://github.com/pat-alt/endogenous-macrodynamics-in-algorithmic-recourse.


The research results include:


Folders with images that went into a) the body of the paper or b) the online companion.
Folders with results (.jls; .csv) for different experiments: a) synthetic data; b) real-world data; and, c) mitigation strategies for both categories of datasets (see paper for details on experiments). Results for all categories are further grouped by dataset.
For each dataset, results include: a) "experiment.jls" files that can be loaded into a Julia session: the loaded Julia objects are structs that contain all settings characterizing a specific experiment. b) "output.csv" files that contain the final experimental outputs: estimated counterfactual evaluation metrics groups by model and counterfactual explainer.