Efficient Partial Dependence Plots with decision trees

Partial Dependence Plots (PDPs) are a standard inspection technique for machine learning models. This post will describe both techniques, and explain why the fast way is… well, faster. We will also see that they are not always equivalent. We will briefly describe partial dependence functions. For a more thorough introduction to PDPs, you can refer to the Bible, or to the Interpretable Machine Learning book.