Paper Type
Complete
Abstract
Spatial risk models are typically developed within individual metropolitan contexts, limiting cross-city comparability and scalability. This study proposes a harmonized multi-city spatial architecture for unified structural risk modeling. We standardize spatial representation using a fixed one-square-mile hexagonal grid, apply globally governed feature construction and normalization, and implement pooled estimation across 35 U.S. metropolitan areas. The dataset includes 11,448 hex cells enriched with 52 structural predictors capturing socioeconomic and built-environment conditions. Structural exposure to gun violence is defined over a three-year window and modeled using logistic regression and ensemble methods. Pooled models achieve AUC values of 0.868-0.877 without city-specific recalibration. Geographically weighted diagnostics indicate limited residual clustering and stable directional effects across contexts. The findings suggest that architectural harmonization can support scalable spatial decision support across heterogeneous urban systems. The study should be interpreted as a proof-of-concept demonstration of architectural feasibility, rather than a comprehensive validation across domains.
Paper Number
1353
Recommended Citation
Satpathy, Asish; Subramanian, Vinay Vikkranth; and Tiwari, Avantika, "A Harmonized Multi-City Architecture for Structural Spatial Risk Modeling" (2026). AMCIS 2026 Proceedings. 9.
https://aisel.aisnet.org/amcis2026/sig_dsa/sig_dsa/9
A Harmonized Multi-City Architecture for Structural Spatial Risk Modeling
Spatial risk models are typically developed within individual metropolitan contexts, limiting cross-city comparability and scalability. This study proposes a harmonized multi-city spatial architecture for unified structural risk modeling. We standardize spatial representation using a fixed one-square-mile hexagonal grid, apply globally governed feature construction and normalization, and implement pooled estimation across 35 U.S. metropolitan areas. The dataset includes 11,448 hex cells enriched with 52 structural predictors capturing socioeconomic and built-environment conditions. Structural exposure to gun violence is defined over a three-year window and modeled using logistic regression and ensemble methods. Pooled models achieve AUC values of 0.868-0.877 without city-specific recalibration. Geographically weighted diagnostics indicate limited residual clustering and stable directional effects across contexts. The findings suggest that architectural harmonization can support scalable spatial decision support across heterogeneous urban systems. The study should be interpreted as a proof-of-concept demonstration of architectural feasibility, rather than a comprehensive validation across domains.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SIG DSA