Modeling Spatial Land Value Dynamics Around Airports with Neural Network

The construction of Dhoho Airport in Kediri Regency has triggered significant spatial and economic transformations in its surrounding areas. This study aims to predict land value in 2024 within a 1-kilometer radius of the airport by employing an Artificial Neural Network (ANN) model, developed using historical land value data from 2022 and 2023. The model integrates spatial and non-spatial predictors, including zoning, road hierarchy, and proximity to infrastructure, and incorporates land cover change analysis from high-resolution satellite imagery. Using a bottleneck architecture and regularization techniques, the ANN model achieved strong predictive performance (RMSE = 37,124.33, MAE = 33,687.45, MAPE = 9.07%, and R2 = 0.9519). SHAP and PFI analyses revealed that distance to collector roads, proximity to the airport, and land zoning are the most influential variables. Spatial analysis of land cover change between 2022 and 2024 indicated that parcels converted from agricultural or vacant land to built-up areas-particularly those located near airport access routes-experienced the highest predicted increases in land value. These results highlight the critical role of accessibility and land use transformation in shaping future land value patterns. The integrated approach highlights the effectiveness of combining machine learning and remote sensing techniques to support urban planning, land valuation, and infrastructure-oriented development policies. The results of this research can be leveraged to inform infrastructure development initiatives that align with the objectives of the Sustainable Development Goals (SDGs), particularly SDG 11 on sustainable cities and communities.