Fair machine learning
summary
Shows a concrete fairness analysis of property tax valuation using tidymodels code and a GitHub repo (github.com/simonpcouch/conf-24) and slides, not abstract ethics. Demonstrates regressivity (predictions pulled to the mean), compares IID residual calibration vs percentage-error parity, and explains how policy choices like a homeowner exemption change which error measure matters. Presents minimal but usable tidymodels fairness functions, extensive contextual documentation, and two applied articles (GPT detector and hospital readmission) so viewers see runnable code and when the tools succeed or fail.
AI-written summary · about
chapters (5)
related talks
19mOpen Source Property Assessment: Tidymodels to Allocate $16B in Property Taxes
35mOpen Source Fairness
42mFairness in Machine Learning: From Theory to Practice
37mKeynote: Ethical Machine Learning: Creating Fair Models in an Unjust World
47mEthical Machine Learning
36mMachine Learning Fairness: Lessons Learned
39mBeyond Bias: Crafting Fairness in Pythonic Machine Learning Systems
28mMeasuring Model Fairness