ModelOriented/DALEX

Health v2 · maintenance only

Adoption not part of the score

1485 stars · 171 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DALEX (moDel Agnostic Language for Exploration and eXplanation) is a library for exploring, explaining, and visualizing the behavior of predictive machine learning models. It wraps any black-box model and provides local and global explainers (variable importance, partial dependence, Shapley values, LIME, Break Down, ceteris paribus profiles), plus fairness auditing and interactive dashboards.

Use cases

• explain predictions from a black-box model like xgboost or keras
• compute feature importance and partial dependence profiles for a model
• compare explanations across models built in different frameworks
• audit a model for fairness and bias
• visualize how a model responds to changes in a single observation
• teach interpretable machine learning concepts
• explore model residuals and performance diagnostics

When to choose

• you need model-agnostic explanations for any predictive model regardless of framework
• you want a comprehensive XAI toolkit with both local and global explainers in R or Python
• you need fairness/bias detection modules alongside explanations
• you want interactive dashboards (Arena) for exploring models

When to avoid

• you need explanations specific to deep learning with gradient-based methods like integrated gradients
• you want a lightweight single-purpose explanation library rather than a full framework
• you need production-scale low-latency explanation serving

Facets

library · maturity active

machine-learning data-visualization nlp machine-learning data-science python xai explainable-ml model-explanations fairness responsible-ai black-box-models predictive-modeling explainable-ai interpretability r

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ModelOriented/DALEX")