Welcome to EthicML’s documentation!

Note

If you are looking for the pre-1.0 documentation, you can find that here: https://wearepal.ai/EthicML-0.x/

Package for evaluating the performance of methods which aim to increase fairness, accountability and/or transparency of machine learning models.

Example

import ethicml as em
import ethicml.data as emda
from ethicml import metrics, models

results = em.evaluate_models(
    datasets=[emda.Adult()],
    inprocess_models=[models.SVM(), models.Kamiran()],
    preprocess_models=[models.Upsampler()],
    metrics=[metrics.Accuracy()],
    per_sens_metrics=[metrics.ProbPos(), metrics.TPR()],
    repeats=5,
)
em.plot_results(results, "Accuracy", "prob_pos_Male_0÷Male_1")

Table of contents

API

• ethicml
  • Cross-validation
  • Model evaluation
  • Parallel evaluation
  • Data preprocessing
  • Utils
  • Visualisation helpers
• ethicml.data
• ethicml.models
  • Algorithms base
  • Inprocess algorithms
  • Postprocess algorithms
  • Preprocess algorithms
• ethicml.metrics

Tutorials

• Migration guide to 1.0
• Runnning experiments on the Adult dataset

Indices and tables

• Index

• Module Index

• Search Page