Iranian Sustainable Building Scientific Association International Journal of Urban Management and Energy Sustainability 2538-1628 7 2 2026 04 01 Explaining a Framework for the Optimal Location of Urban Neighborhood Facilities Using an Artificial Neural Network in a GIS Environment (Case Study: Eastern Zone of Tabriz Metropolis) 66 79 737855 10.22034/ijumes.2026.2091576.1374 EN Ayda Mahmoudzadeh Department of Architecture and Urban Planning, Ta.C., Islamic Azad University, Tabriz, Iran 0000-0003-3945-0992 Arash Saghafi Asl Department of Architecture and Urban Planning, Ta.C., Islamic Azad University, Tabriz, Iran 0000-0001-5627-2353 Habib Shahhoseini Department of Architecture and Urban Planning, Ta.C., Islamic Azad University, Tabriz, Iran 0000-0002-1385-1440 Masoud Haghlesan Department of Architecture and Urban Planning, Ilk.C., Islamic Azad University, Ilkhchi, Iran. 0000-0001-9826-0086 Journal Article 2026 04 14 The optimal location of neighbourhood-scale urban facilities is a foundational determinant of spatial justice, service accessibility, and the long-term liveability of cities, and in large, morphologically heterogeneous metropolises such as Tabriz it becomes a problem of considerable analytical complexity. The central problem addressed in this study is that prevailing location-allocation practice in Iranian metropolitan areas relies on linear, weight-based multi-criteria techniques that cannot represent the non-linear, high-dimensional interactions among the spatial, physical, and socio-economic determinants that actually govern urban suitability, producing inequitable and operationally fragile facility distributions. The objective of this research is therefore to develop, train, and validate an operational framework for the optimal location of neighbourhood facilities by integrating a multilayer-perceptron Artificial Neural Network with a Geographic Information System spatial database, using the eastern zone of Tabriz as a case study. Findings indicate that the trained network achieved a coefficient of determination of 0.974 and a root-mean-square error of 0.040 on the validation set, that residential fabric density, land-use compatibility, and distance to existing services were the most influential criteria, and that approximately 38.7 per cent of the study area was classified as suitable or highly suitable for new facility development, with recommended sites concentrated in identifiable high-suitability corridors. The study concludes that ANN–GIS integration constitutes a methodologically robust, interpretable, and operationally transferable instrument for evidence-based facility location in Iranian metropolitan contexts, materially outperforming conventional linear weighting. https://www.ijumes.com/article_737855_de82a5d7f0766a76544f5499bf5ee7d7.pdf