Interpreting Models via Single Tree Approximation

Yichen Zhou, Giles Hooker

We propose a procedure to build a decision tree which approximates the performance of complex machine learning models. This single approximation tree can be used to interpret and s…