Comparison Of Three Classification Methods For Feature Selection In Diabetes Data

Abstract

This paper describes a focus on predicting and classifying people suffering from diabetes using 3 classification techniques. It aims to provide a correct diagnosis at the right time to prevent fatal outcomes. The diabetes is a significant global health concern and is one of the leading causes of death worldwide. The paper employs various techniques to predict and classify individuals at risk of diabetes. These techniques include the filter method, wrapper method, and genetic algorithm method. These methods are often used for feature selection and model building in machine learning. The study suggests that while heuristic methods may not be as accurate as classification methods, the results are satisfactory. The AUC (Area Under the Curve) value reached 80% in a hybrid combination of the genetic algorithm (GA) method with the GSA (Gravitational Search Algorithm). The proposed system can easily distinguish between healthy and unhealthy individuals, which is essential for early intervention and treatment. The algorithms used in the study have a running times and memory usage by 98.75%. The combination of different algorithms, such as GA and GSA, can help doctors diagnose sick patients efficiently and on time.  The main objective, the paper appears to offer a method for predicting and classifying diabetes, highlighting the importance of early diagnosis and the potential benefits of using a combination of different algorithms to improve efficiency and accuracy. 80% of AUC value suggests that the model's performance is quite promising in distinguishing between healthy and diabetic individuals. However, it's important to note that the effectiveness of such models can vary depending on the quality and quantity of data used in training and testing.