Predicting Urban Expansion in Zacatecas-Guadalupe cities: A Support Vector Machine Approach Enhanced by SHAP Values

Abstract

Cities are centers that generate employment, innovation, and improve the quality of life, basic services, and housing. Rapid and unplanned expansion brings with it undesirable consequences for social and economic development. In Mexico, three out of four people live in a city. For this reason. the aim in this paper is to model and predict urban expansion in Zacatecas-Guadalupe cities (Mexico) using support vector machines, and thus carry out a better planning. In order to achieve this objective, land use and land cover maps corresponding to the period 2000–2020 were used, as well as the inclusion of socioeconomic, topographic, and cultural attribute variables. A soft SVM model was developed with a training accuracy of 92.4%, a validation accuracy of 93% and an F1-Score of 86.3%. In the obtained results, it can be observed that the proximity to already urbanized areas and the type of land use have a high influence on urbanization