Enhanced Local Explainability and Trust Scores with Random Forest Proximities

no code implementations • 19 Oct 2023 • Joshua Rosaler, Dhruv Desai, Bhaskarjit Sarmah, Dimitrios Vamvourellis, Deran Onay, Dhagash Mehta, Stefano Pasquali

We initiate a novel approach to explain the predictions and out of sample performance of random forest (RF) regression and classification models by exploiting the fact that any RF can be mathematically formulated as an adaptive weighted K nearest-neighbors model.

Prediction regression