Explainable Predictive Modeling of Obesity Using Machine Learning

This project aims to predict obesity risk using machine learning algorithms, focusing on model accuracy and interpretability. Various models like XGBoost and Stacking Classifiers are employed to classify individuals based on features such as age, family history, and lifestyle factors. The system incorporates Explainable AI (XAI), specifically LIME, to provide explanations for predictions, helping users understand the reasoning behind the results. The project compares the performance of Naive Bayes and Support Vector Machine (SVM) with the proposed models, offering improved accuracy and explainability. The goal is to make the prediction process transparent and reliable for healthcare professionals and researchers, enhancing decision-making in obesity risk management. The web-based system allows users to upload data, get predictions, and explore model explanations, creating an accessible tool for obesity prediction and intervention.

Keywords: Obesity prediction, machine learning, stacking classifiers, LIME, CatBoost, LightGBM, Random Forest, feature selection, SMOTE, explainability.