Open Source Property Assessment: Tidymodels to Allocate $16B in Property Taxes
summary
The Cook County Assessor's Office replaces a 50,000‑line SPSS regime with R and tidymodels plus LightGBM, publishing the full pipeline and data in an open-source repo. Models since 2020 value 1.5 million properties (≈$400B) countywide, cut training time by roughly two‑thirds via early stopping, and shift assessments so lower‑value homes are less overassessed while higher‑value homes rise toward market ratios, improving measured accuracy and equity.
AI-written summary · about
chapters (7)
- 0:00Property tax challenges
- 4:04Tax divide and flawed valuations
- 5:49Closed to open source modeling
- 8:46LightGBM model with tidymodels
- 12:44Results and impact
- 14:54Open data and reporting
- 15:58Lessons and call to action
related talks
20mFair machine learning
38mThe Journey Through Real Estate Price Model Development
57mGood Practices for Applied Machine Learning- 57mGood Practices for Applied Machine Learning
37mAI for Good- Then and Now
24mState of the Art Open Source Software for AI Fairnessevaluation
23mTotal Tidy Tuning Techniques
15mThe tidysynthesis R package