Improving ML Models for Obesity Prediction Using GANs and Social Physical Activity Data GAN ler ile Sosyal Fiziksel Aktivite Verilerine Dayali Obezite Tahmininde ML Modellerinin Geli stirilmesi
33rd IEEE Conference on Signal Processing and Communications Applications, SIU 2025, İstanbul, Türkiye, 25 - 28 Haziran 2025, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu66497.2025.11112032
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Generative adversarial networks, Machine Learning, Obesity prediction, Synthetic data generation
- Maltepe Üniversitesi Adresli: Evet
Özet
Obesity is a growing health concern, making early detection and prevention crucial. While Machine Learning (ML) models perform well, some, like Artificial Neural Networks (ANNs), require more data. Generative Adversarial Networks (GANs) can address this by generating synthetic data. This study explores the impact of synthetic data on obesity prediction models using online survey data. Two approaches were tested: (1) training a plain GAN to generate 15000 synthetic samples and (2) using a Conditional Tabular GAN (CTGAN) to create another 15000 samples. Experimental results show that both methods improved model performance across precision, accuracy, recall, and F1-score while reducing data imbalances.