Link Search Menu Expand Document
Papers Video GitHub EXPEKCTATION About

Random Forest Explanations

Let $f$ be a Boolean function represented by a random forest $RF$ and $x$ be an instance. $RF$ is an ensemble $RF = {T_1,\cdots,T_m}$, where each $T_i$ is a decision tree, and such that the prediction $f(x)$ is given by:

\[\begin{align*} f(x) = \begin{cases} 1 & \mbox{ if } \frac{1}{m}\sum_{i=1}^m T_i(x) > \frac{1}{2} \\ 0 & \mbox{ otherwise.} \end{cases} \end{align*}\]

For explanations computation, PyXAI supports only the random forests dedicated to binary classification. Algorithms designed for the calculation of explanations suited to multi-class random forests are under development.


Table of contents