Comparative Analysis of Machine Learning and Deep Learning in Predicting Multiclass Obesity Levels

In recent years, obesity has grown exponentially. Obesity or overweight is a physical condition where there is an excessive accumulation of fat in the body, disrupting one's health. Society tends to feign ignorance to their unhealthy lifestyles which leads to many health problems including obesity. It is a type of disease that can cause many chronic diseases, such as heart disease, diabetes, and much more. Therefore, we propose the idea of using AI to build a system that will serve as preventive measures which serve as the purpose of this research. Meaning that identifying obesity risk will be much easier and will help raise people awareness. Another aim of this paper is not only will readers be able to learn about obesity and its causing factors, but readers will also gain in-depth knowledge about machine learning and deep learning algorithms. Diving deep into the AI algorithms especially reading related works, we found interesting research by Elias Rodriguez et al., where machine learning performs better than deep learning in terms of KNN and MLP algorithms. This piques our interest. Therefore, this paper will further study this phenomenon. For this research, implementation of machine learning and deep learning algorithms widely used in prediction is involved. The algorithms used are K-Nearest Neighbor (KNN) and Multi-Layer Perceptron (MLP). The experiment shows that during certain circumstances MLP has achieved above 89% accuracy. This shows that the model can receive the dataset and predict obesity without any issue. Furthermore, the model can now be used as a preventive measure (early detection) against obesity or overweight.