Authors - Jonalyn Joy B. Labayne, Joey Aviles, Ronald Cordova Abstract - This empirical study examines regional and demographic disparities in BMI and Waist-Hip Ratio (WHR) as indicators of nutritional and cardiometabolic health in the Philippines. Using the 2013 FNRI National Nutrition Survey dataset (n = 69,505 adults aged 20 years and above), data were processed with Apache Spark for distributed handling of large-scale heterogeneous records. A Random Forest classifier was trained with 10-fold stratified cross-validation and inverse class-weighting to mitigate severe class imbalance. The model achieved an accuracy of 0.81, macro-F1 score of 0.67, and area under the precision-recall curve (AUCPR) of 0.75. These FNRI-specific results are discussed in the context of existing literature. Genc and Arıcan (2025) compared eight machine learning algorithms on a Latin American obesity dataset (n = 2,111), excluding height and weight variables; Random Forest achieved the highest ROC AUC of 0.98 and macro-F1 of 0.87 in that study. The inclusion of WHR alongside BMI in the FNRI analysis provides enhanced cardiometabolic risk stratification. The findings underscore the value of ensemble methods in future Philippine research to better detect minority classes and support regionally targeted public health interventions in resource-limited settings.
Open Zoom