Estimation of obesity levels based on computational intelligence

Abstract

Obesity is a worldwide disease that affects people of all ages and gender; in consequence, researchers have made great efforts to identify factors that cause it early. In this study, an intelligent method is created, based on supervised and unsupervised techniques of data mining such as Simple K-Means, Decision Trees (DT), and Support Vector Machines (SVM) to detect obesity levels and help people and health professionals to have a healthier lifestyle against this global epidemic. In this research the primary source of collection was from students 18 and 25 years old at institutions in the countries of Colombia, Mexico, and Peru. The study takes a dataset relating to the main causes of obesity, based on the aim to reference high caloric intake, a decrease of energy expenditure due to the lack of physical activity, alimentary disorders, genetics, socioeconomic factors, and/or anxiety and depression. In the selected dataset, 178 students participated in the study, 81 male and 97 female. Using algorithms including Decision Tree, Support Vector Machine (SVM), and Simple K-Means, the results show a relevant tool to perform a comparative analysis among the mentioned algorithms.

Keywords

• Obesity
• Simple k-means
• Decision trees
• Support vector machines