Feature importance scores are a collection of methods all used to answer one question: which machine learning model features have contributed the most to predictions in general? Amongst all these methods, permutation feature importance is the most popular. This is due to it’s intuitive calculation and because it can be applied to any machine learning model. Understanding PFI is also an important first step in understanding more complex explainable AI methods like SHAP, LIME and PDPs.

In this video, we’re going to gain a deep understanding of PFI. To do this, we will use Python to calculate importance scores from scratch. We will also discuss the logic behind the method including why we permute, repeat and which metric to use. Really, we will be focussing on the P in PFI. Permutation is an important part of many model-agnostic methods. Taking time to understand this approach and its limitations will make those methods easier to understand.

🚀 Free Course 🚀
Signup here: https://mailchi.mp/40909011987b/signup
XAI course: https://adataodyssey.com/courses/xai-...
SHAP course: https://adataodyssey.com/courses/shap...

🚀 Companion article with link to CODE (no-paywall link): 🚀
https://medium.com/towards-data-scien...

🚀 Useful playlists 🚀
XAI:    • Explainable AI (XAI)  
SHAP:    • SHAP  
Algorithm fairness:    • Algorithm Fairness  

🚀 Get in touch 🚀
Medium:   / conorosullyds  
Threads: https://www.threads.net/@conorosullyds
Twitter:   / conorosullyds  
Website: https://adataodyssey.com/

🚀 Chapters 🚀
00:00 Introduction
01:15 PFI from scratch
06:54 The logic behind PFI
10:02 The limitations of permutation