Opening the "black box" of machine learning models has been huge in not only understanding the models we create, but then also communicating the insights to others. As I encounter different use cases for explainable AI, I'm distilling the insights into manageable chunks and sharing them openly.
- Demonstrate a way to explore the explainability of multi-output regrssion models with SHAP.
- View ipynb here (Recommend downloading and running the entire notebook in Google Colab)
SHAP Values for Multi-Output Regression Models
Apply Shapely values to a multi-output regression model to explore how the features effects each of the outputs/labels independently.