Keyword: Artificial Inteligence, Spatial Database, Spatial Modeling, Web Developement, Data Analysis
2022
Research
Optimizing Housing Subsidy Allocation Using Predictive Modeling applies logistic regression and cost-benefit analysis to improve how limited housing subsidies are distributed. Rather than allocating resources randomly, the study identifies individuals most likely to apply for and receive subsidies, helping policymakers maximize both efficiency and impact.
Housing assistance programs often face funding shortages and uneven allocation outcomes. Many eligible households remain unreached, while others receive aid inefficiently. This project explores how predictive modeling can strategically target high-probability recipients, ensuring limited resources achieve greater policy effectiveness while minimizing administrative costs.
Using data from 4,119 applicants, the study built logistic regression models to predict housing subsidy uptake. Two versions—a full and a feature-engineered model—were tested and evaluated with ROC and AUC metrics, with the engineered model performing best (AUC 0.78). A cost-benefit analysis then optimized the decision threshold at 0.17, increasing projected revenue by 181% over the standard 0.50 cutoff and showing the value of data-driven policy targeting.
Conducted in 2022 as part of the Master of Urban Spatial Analytics (MUSA) program at the University of Pennsylvania, this project was completed under the guidance of Professors Elizabeth Delmelle, Michael Fichman, and Matt Harris.
The optimized model achieved nearly 90% higher sensitivity than a random allocation baseline, accurately identifying recipients most likely to accept housing subsidies. By integrating predictive modeling with financial optimization, the analysis demonstrated that strategic targeting could significantly improve policy performance and fiscal efficiency. The results highlight how quantitative methods can enhance equity and impact in public resource allocation.