Obesity Prediction using Bayesian Optimized Gradient Boosted Trees

Edric Castel Hao, Madelein Villegas, Dino Dominic F. Ligutan, Jose Martin Maningo

Abstract: Obesity is a growing concern worldwide, and its prediction is crucial for preventing and managing related health problems. In this study, we aim to develop a reliable and accurate model for predicting obesity based on various demographic and lifestyle factors. Prior studies have been trying to combat obesity by understanding the underlying causes and developing interventions to treat and prevent the disease. The model considers data such as age, gender, aspects of physical activity, and dietary habits from the publicly available dataset in the University of California Irvine (UCI) Machine Learning Repository to make predictions. We used different machine learning algorithms to analyze the data and evaluate the performance of the model. Our results showed that the model had a high accuracy rate of 99.84% from the LightGBM model, which indicates its potential for practical use. Nested Stratified Cross-Validation (CV) was used to confirm the results of the model prediction. Furthermore, the model provides valuable insights into the factors that contribute to obesity, which can be used to inform public health policies and interventions. In conclusion, this study has important implications for the prevention and management of obesity and highlights the importance of considering demographic and lifestyle factors in obesity prediction.