Geospatial Artificial Intelligence-based Approach for Predicting Slum Growth, a Case in Iraq

Sarhan, Hussein Riyadh; Mousa, Yousif A.; Hanoon, Sarah Abdulkareem; SACHIT, MOURTADHA SARHAN

doi:10.24271/psr.2025.527086.2166

Geospatial Artificial Intelligence-based Approach for Predicting Slum Growth, a Case in Iraq

Document Type : Original Article

Authors

Hussein Riyadh Sarhan ¹

Yousif A. Mousa ²

Sarah Abdulkareem Hanoon

MOURTADHA SARHAN SACHIT

¹ Department of Surveying and Geomatics Engineering, University of Thi-Qar, Nasiriyah 64001, Iraq.

² Civil engineering department, college of engineering, Al-Muthanna University, Samawah 66001, Iraq.

10.24271/psr.2025.527086.2166

Abstract

The spread of slums poses a major challenge to urban planning, necessitating the development of effective predictive models to mitigate its impact on sustainable development. This study aims to spatially predict areas at risk of slum expansion in Al-Bathaa City, Iraq. To this end, conditioning factors were identified and developed using Geospatial Information Systems (GIS), including proximity to roads, agricultural lands, public services, neglected areas, and water sources. Three Artificial Intelligence (AI) algorithms were examined: Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM). The models' performance was evaluated using several metrics, such as classification accuracy (CA), F1-score, recall, and precision. The most accurate model was then utilized to predict the spatial expansion of slums. The results revealed the superiority of the RF algorithm, which achieved a CA of 0.949. The slum risk index indicated that more than 12% (~600,000 m²) of the study area is at high risk of slum expansion in the near future. The developed slum expansion index will assist decision-makers in responding quickly and devising appropriate plans to prevent slum expansion, thereby promoting urban sustainability and effective land-use planning.

Keywords

Slum

GIS

GeoAI

Remote sensing