Integrating artificial neural networks and neighborhood analysis for enhanced urban growth prediction depending on planned future development areas: A case study of Ar-Ramtha, Jordan

Abstract

Artificial intelligence (AI) is revolutionizing urban planning by enhancing the simulation of urban growth and land-use changes. Most AI models often fail to predict how proposed development areas will shape future expansion, leading to misaligned planning and urban sprawl. This study addresses these gaps by integrating master plans and spatial policies into an AI-driven urban growth model for Ar-Ramtha. A hybrid model combining artificial neural networks (ANN) and neighborhood analysis was developed to assess the influence of proposed development areas. Historical data (1990–2005) was used for training and validation, while 2005–2020 data was used for testing. Under-sampling techniques addressed data imbalance, and the PLADJ matrix measured urban growth patterns. Results indicate that incorporating neighborhood analysis and development areas improves model accuracy, achieving a validation accuracy of 92.9 ​%, an F1-micro of 91.0 ​%, and a Kappa of 82.1 ​%. Testing performance remained strong, with an F1-micro of 84.0 ​% and an accuracy of 91.2 ​%. Simulations suggest Ar-Ramtha is becoming more compact, with PLADJ values rising from 80.67 ​% in 2020 to 92.7 ​% by 2035. However, potential sprawl risks remain due to increased road paving and rising land prices. The study recommends defining development areas based on AI model predictions beyond current planning boundaries and adopting a ring road strategy to promote compact urban growth. These findings contribute to bridging gaps in AI-based urban modeling by improving the integration of planning policies, offering a robust tool for sustainable urban development.