Date : 23/06/2025
Hour: 20:30
Student Name & ID: Abbas Ali Elazhari , 245 105 404
Supervisor: Prof. Dr. Ömer KARAL
Topic: Obesity Prediction Using Machine Learning: A BMI Classification Approach with Anthropometric and DXI-derived Features
Link or Room: https://meet.google.com/hhb-bmhu-ggb
Meeting Link(Online) : Link will be given by the Supervisor later on.
OR
Room Number(Face to Face)
Abstract:
This study explores the application of multiple machine learning classifiers to predict Body Mass Index (BMI) categories among children and adolescents using anthropometric and DXI-derived features. The dataset presented significant challenges due to missing values and severe class imbalance. Rigorous preprocessing was applied, including data cleaning, normalization, and class balancing through SMOTE and manual resampling. A range of classifiers including Random Forest, SVM, XGBoost, LightGBM, CatBoost, MLP Neural Network, and Logistic Regression were trained using GridSearchCV with 5-fold cross-validation. Ensemble methods, particularly Random Forest and LightGBM, achieved the highest accuracy, reaching 99.5%. Experimental results validate the effectiveness of preprocessing and model tuning in enhancing performance for health-related classification tasks.
