Forest Guided Smoothing
This work addresses the need for interpretable machine learning methods in domains like healthcare, though it appears incremental as it builds on existing random forest techniques.
The paper tackles the problem of making random forests more interpretable and usable for tasks like bias correction and confidence intervals by introducing a family of local smoothers with adaptive bandwidth matrices derived from forest outputs, resulting in a method that retains flexibility while enabling linear interpretability and application to previously intractable tasks.
We use the output of a random forest to define a family of local smoothers with spatially adaptive bandwidth matrices. The smoother inherits the flexibility of the original forest but, since it is a simple, linear smoother, it is very interpretable and it can be used for tasks that would be intractable for the original forest. This includes bias correction, confidence intervals, assessing variable importance and methods for exploring the structure of the forest. We illustrate the method on some synthetic examples and on data related to Covid-19.