Using Decision Trees, Logistic Regression and Random Forest to Predict Poverty Risk in Thailand

The study explores the potential of geospatial AI and the Rajabhat dataset, which includes information on 11,240 individuals from Pathum Thani and Sakaew provinces, to predict and improve progress towards Sustainable Development Goal (SDG) achievement in Thailand. It utilises AI models such as decision trees, random forests, logistic regression, and support vector machines (SVM) to analyse geospatial features and forecast progress towards SDG targets. The findings show that logistic regression and random forest models demonstrate high accuracy levels of over 95% in predicting SDG indicator values. The study also identifies critical geographical factors that significantly impact SDG outcomes. It creates maps to visualise predicted SDG progress at different geographical scales, helping to identify areas that may need targeted interventions. The research concludes by emphasising the effectiveness of AI and geospatial data in supporting evidence-based decision-making for sustainable development, particularly in guiding resource allocation and policy adjustments to accelerate SDG achievement in Thailand.