Compute SHAP values to explain model predictions and feature contributions.
npx skills add https://github.com/HaykTarkhanyan/dst_research --skill shap-hayktarkhanyan
Please help me install this Agent Skill. Skill: shap Source: https://github.com/HaykTarkhanyan/dst_research/tree/main/.claude/skills/shap Command: npx skills add https://github.com/HaykTarkhanyan/dst_research --skill shap-hayktarkhanyan
SHAP-based explanations address the challenge of understanding and communicating how machine learning models arrive at their predictions, turning opaque black-box decisions into transparent, feature-level attributions that stakeholders can trust.
Train your model, create a SHAP explainer, and compute SHAP values for your data.
High-intent search queries and answers about installing and using this skill.
Yes, SHAP interpretations support tree-based, deep learning, linear, and black-box models. The skill selects the appropriate explainer for each model type and handles background data to generate accurate feature attributions.
SHAP visualizations help you communicate model explanations by generating waterfall, beeswarm, bar, scatter, force, and heatmap plots. These visualizations translate opaque black-box decisions into transparent, feature-level attributions.
To compute SHAP values, you first train your model, create a SHAP explainer, and compute the values for your data. This process yields numerical attributions that support model comparison and production deployment.
For model interpretability, you can generate waterfall, beeswarm, bar, scatter, force, and heatmap visualizations. These plots communicate feature contributions and turn opaque black-box decisions into transparent attributions.
ManfronEnricoCompute SHAP values and visualizations for tree, deep, and linear models.
Weiwei-MaoCompute SHAP feature attributions and generate explanation plots for machine learning models.
CRAG666Compute SHAP values and generate global and local explanation plots.
tondevrelAnalyze scikit-learn model feature importance with SHAP, LIME, and partial dependence plots.
