Start Date
4-12-2024 1:15 PM
End Date
4-12-2024 2:45 PM
Description
Cook County's 1.2m residential properties are revalued using a nonlinear tree-based model. This talk will combine visualization and analytics to share some lessons learned about model comparison and an algorithm for model transparency. First, we'll show how we rigorously conduct different model tests and visualize IAAO statistics for easy comparison. Once we have a model we trust, our goal is to earn the trust of the public who are becoming increasingly aware of AI tools and their pitfalls. We'll discuss a novel algorithm that extracts each subject property's sale comps used by the nonlinear model and demonstrate how we share this data with the goal of improving transparency and trust in government.
Publication Date
December 2024
Recommended Citation
Jardine, Nicole PhD, "Measuring Impact: New Statistical Methods for Assessment Accuracy and Transparency" (2024). International Research Symposium. 5.
https://researchexchange.iaao.org/irs/irs24/sessions/5
Measuring Impact: New Statistical Methods for Assessment Accuracy and Transparency
Cook County's 1.2m residential properties are revalued using a nonlinear tree-based model. This talk will combine visualization and analytics to share some lessons learned about model comparison and an algorithm for model transparency. First, we'll show how we rigorously conduct different model tests and visualize IAAO statistics for easy comparison. Once we have a model we trust, our goal is to earn the trust of the public who are becoming increasingly aware of AI tools and their pitfalls. We'll discuss a novel algorithm that extracts each subject property's sale comps used by the nonlinear model and demonstrate how we share this data with the goal of improving transparency and trust in government.
