Azahra, Renaya Aviary (2025) COMPARISON OF NAIVE BAYES AND KOLMOGOROV-ARNOLD NETWORKS (KANs) METHODS IN THE CLASSIFYING OF OBESITY LIKELIHOOD LEVELS. S1 Undergraduate thesis, Universitas Darussalam Gontor.
![[img]](https://repo.unida.gontor.ac.id/style/images/fileicons/text.png)
Abstract
Obesity is one of the world’s most serious health problems, and its prevalence is increasing every year. Several other health problems occur after someone is obese. Predicting the possibility of obesity by detecting it as early as possible is a very important step in reducing the risk of related health complications. Obesity can be detected immediately by looking at several factors that trigger obesity. Some of these factors are height, weight, and body mass index. These three factors are the primary triggers for obesity. This study examines the performance of two classification methods, Naïve Bayes and Kolmogorov-Arnold Networks(KANs). The analysis was performed by comparing the accuracy of each method. The data is processed by cleaning, testing, and evaluation. The results show that the Naïve Bayes method is superior to Kolmogorov-Arnold Networks. Namely 96% for Naïve Bayes and 71% for Kolmogorov-Arnold Networks. This indicates that Naïve Bayes is more effective on the data examined. Naïve Bayes is a reasonably practical probability theory even though it uses simple data. On the other hand, Kolmogorov-Arnold Networks shows limitations in the data set because it requires complex data, characterized by many variables and a more significant amount of data.
Statistics Downloads of this Document
 |
View Item |