liveBook · Manning

Shapley value

As explained in Build AI Drug Discovery Pipelines

A Shapley value is a cooperative-game-theoretic attribution that assigns an average marginal contribution to an individual feature across coalitions.

A Shapley value attributes a model prediction to individual descriptors by averaging each descriptor’s marginal contribution across possible coalitions. In molecular modeling, descriptors act as coalition members and the model output is the coalition’s value. section 6.4.3

A practical attribution workflow is:

  • Choose a molecule and its descriptors, such as caffeine’s hydrogen, carbon, oxygen, and nitrogen counts. section 6.4.3

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Explains coalition-based marginal contributions and reaches practical SHAP computation and visualization.
Examples
Uses caffeine descriptors, RNA-binder data, XGBoost, TreeExplainer, beeswarm plots, cohort bars, scatter plots, and waterfalls.
What it assumes
Starts from descriptors, coalitions, and model outputs while defining cooperative-game intuition before applying it.
What you can do after
Apply it via SHAP to rank descriptors globally and explain predictions for individual molecules.
How it connects
Connects cooperative game theory, surrogate models, molecular descriptors, XGBoost, and model interpretation.

Also explained in

Compare how 3 books explain Shapley value →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.