Genetic Algorithm Based Feature Subset Selection for Fetal State Classification

Abstract

Abstract-Huge amount of data are available in the field of medicine which are used for diagnosing the diseases by analysing them. Presently, prediction of diseases are made easier and accurate by employing various data mining techniques to extract information from these medical data. This paper presents an improved method of classifying the cardiotocogram data using Multiclass Support Vector Machine (MSVM) through an optimized feature subset produced by Genetic Algorithm (GA). Various performance metrics have been evaluated and the experimental results exhibit improved performance when using optimized feature set comparing to the full feature set.