Abstract:

Although obesity is more common in developed countries, it is also a disease in developing countries that negatively affects people's quality of life. Although there are several factors that trigger obesity, some of the most concrete examples of these factors can be mentioned: sedentary lifestyle, unbalanced diet. Obesity can show up in patients at different levels. Each of these levels is important for treatment, so the diagnosis of obesity must be made early. With this in mind, an autonomous system was developed that can assist experts in decision making. In the study, a machine learning-based approach was proposed to classify obese patients according to their degree of obesity. Feature selection was made with Neighborhood Component Analysis (NCA) method on 2111 patient data with 16 features from UCI machine learning repository, and features were classified with Extreme Gradient Augmentation (XGBoost) and Decision Tree algorithms. When the results were examined, a 1% increase in accuracy rates was observed after feature selection for both algorithms. Since the system works according to the purpose, it is expected that a study will be conducted to help experts with the optimal number of features for predicting the degree of obesity.

Keywords: