Comparative Analysis of Linear and Polynomial Regression for Predicting Body Mass Index

Body Mass Index (BMI) is a widely used metric for assessing obesity-related health risks and guiding public health interventions. This study evaluates and compares the effectiveness of linear and polynomial regression models in predicting BMI based on gender, macronutrient intake, and anthropometric data from an adult population in Montenegro. The analysis incorporates feature engineering, introducing variables such as carbohydrate intake per kilogram of body weight, water intake per kilogram, and waist-to-height ratio to enhance predictive accuracy. Model performance is assessed using multiple evaluation metrics, including predictive error and explanatory power, with results indicating that polynomial regression provides only marginal improvements over linear regression. These findings reinforce the viability of linear regression as an interpretable and efficient approach for BMI estimation. The study highlights the need for future research to explore advanced machine learning techniques to better capture the complex, nonlinear relationships influencing BMI and improve predictive accuracy in public health applications.