Enhancing Obesity Detection through Machine Learning Algorithms
6th International Conference on Innovative Computing, ICIC 2025, Lahore, Pakistan, 10 - 11 Aralık 2025, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/icic68258.2025.11412977
- Basıldığı Şehir: Lahore
- Basıldığı Ülke: Pakistan
- Anahtar Kelimeler: Body Mass Index (BMI), linear Regression, Machine Learning (ML), Obesity Prediction, Random Forest
- İstanbul Gelişim Üniversitesi Adresli: Hayır
Özet
Obesity is major global public health concern associated with increased risk of chronic diseases, cardiovascular disorders, and metabolic syndromes. Early and accurate detection of obesity can support preventive interventions and more effective clinical decision-making. This study presents a comprehensive benchmark analysis of multiple machine learning (ML) algorithms for obesity detection using a balanced and well-defined dataset comprising 3,114 samples and 19 physiological, demographic, and lifestyle-based features. Unlike prior studies that rely on limited classifiers, unbalanced datasets, and inconsistently defined target labels, this work establishes a reproducible evaluation framework incorporating systematic preprocessing, missing values imputation, feature encoding, hyperparameter tuning, and 10-fold cross-validation. 10 ML Algorithms-Random Forest (RF), Support vector Machine (SVM), MLP, Fuzzy K-NN, FURIA, Linear Regression (LR), Naïve Bayes (NB), DT, RS and RT are compared using accuracy, precision, recall and Fl-Score. Results demonstrate that RM achieves the highest accuracy (97.88%) and best overall balance across metrics, followed by LR (96.72%), and MLP (93.75%). The Study highlights interpretability through feature importance analysis and clarifies algorithmic tradeoffs. By presenting a robust and transparent benchmarking methodology, this work provides a strong foundation for future predictive healthcare system and demonstrate the potential of ML-driven obesity risk assessment.