Comparison of Built-Up Expansion Prediction Models for Magelang City and Its Peri Urban Areas: CA-Logistic Regression, CA-SVM, CA-MLP, and CA-RF

Abstract

Monitoring and prediction of built-up land expansion is necessary to support spatial planning. This study predicts built-up land expansion using CA-Markov integrated with four machine learning models, including logistic regression, Support Vector Machine (SVM), Multilayer Perceptron (MLP), and Random Forest (RF). These models were applied to Magelang City and its peri-urban areas from 2013 to 2025. During 2013 to 2025, Magelang City and its peri-urban areas experienced LULC changes. Residential buildings, non-residential buildings, agriculture land, and agroforestry land increased in area over time. Meanwhile, bare land and shrubland decreased in area over time. The results also show that there was an expansion of built-up land in Magelang City and its peri-urban areas from 2013 to 2025, from 2,674.35 ha to 3,483.81 ha.  The 2013 and 2019 LULC maps were used as input in predicting built-up land expansion in 2025. The model was trained using LULC change in 2013 to 2019 and selected driving factors. The driving factor with the most significant influence was the distance to existing built-up land. The results showed that the SVM model achieved the highest accuracy, with an overall accuracy of 98.4% and a Figure of Merit (FOM) of 15.3%. This indicates that the SVM model is a most suitable model for predicting built-up land expansion in Magelang City and its peri-urban areas.