Getting started

Install

Install from TestPyPI:

pip install -i https://test.pypi.org/simple/ \
    --extra-index-url https://pypi.org/simple \
    recourse-bench

Or install from a source checkout:

pip install -r requirements.txt

Minimal config

RecourseBench experiments are configured as data. A config chooses a dataset, preprocessing, target model, recourse method, and evaluation metrics.

name: toy_linear_wachter
seed: 7
dataset:
  name: toydata
preprocess:
  - name: scale
    scaling: standardize
    range: true
  - name: split
    split: 0.25
model:
  name: linear
  device: cpu
method:
  name: wachter
  device: cpu
  desired_class: 1
evaluation:
  - name: validity
  - name: distance

Run

From the command line:

recourse-bench -p experiment/toy/smoke_config.yaml

From Python:

import yaml
from pathlib import Path
import recourse_bench as rb

config = yaml.safe_load(Path("experiment/toy/smoke_config.yaml").read_text())
metrics = rb.run(config)
print(metrics.to_string(index=False))

Use the registry helpers to discover valid component names:

rb.list_datasets()
rb.list_models()
rb.list_methods()
rb.list_evaluations()

rb.run(config) trains the target model, fits the recourse method, generates counterfactuals, and returns a one-row metrics dataframe.