Childhood obesity is a significant public health issue in Mexico, affecting 38.2% of children. This study aimed to explore the relationship between lifestyle habits, food environments, and nutritional status in 230 schoolchildren from Zapotlán el Grande, Jalisco, using machine learning techniques. Nutritional status was assessed using anthropometric measurements (weight, height, Body Mass Index (BMI), waist circumference, and waist-to-height ratio [WHtR]). For its part, dietary and physical activity habits were assessed using the validated Mexican questionnaire (HLHDPAQ). Decision tree models (J48) were fitted to characterize patterns that identify children’s nutritional status, reinforcing findings from conventional statistics. This methodology enabled us to determine thresholds for obesity severity based on the CDC’s BMI percentiles. Another model fit highlights the importance of WHtR as a strong predictor of obesity, as well as its interaction with BMI and specific habit responses. The last model showed that body fat percentage is a crucial variable for the accurate classification of obesity. Across all models, poor dietary habits and physical inactivity were consistently associated with increased obesity risk. These results emphasize the importance of using multiple indicators, such as WHtR and body fat percentage, for a more comprehensive assessment of childhood obesity. Furthermore, they underscore the importance of creating salutogenic environments—those that promote health and well-being—by regulating school food environments and increasing access to public recreational spaces. Decision trees offer a powerful tool for identifying critical risk factors, guiding personalized interventions, and supporting public health policies to prevent childhood obesity.