Classification of delivery type of pregnant women using support vector machine

Informatics Department, Sanata Dharma University, Paingan Maguwoharjo Depok Sleman Yogyakarta, Indonesia

^* Corresponding author: chelseayubela@putranirmala.sch.id

Abstract

One of the ways to reduce maternal mortality is by diagnosing childbirth to find out whether a mother will give birth normally or not so that appropriate treatment can be done. This study aims to improve maternal safety and health by classifying delivery type of pregnant women, either Caesarean or normal types, using the Support Vector Machine method. The dataset used in this study was taken from a hospital in 2020. It consists of 25 attributes and 302 records that include information about the health conditions of pregnant women and babies. Several experiments were performed towards the dataset with and without balancing. Three types of SVM kernels, namely Linear, RBF, and Polynomial kernels, were then implemented to classify the dataset using several variations of parameters of C, gamma, and degree. The validation was performed using several k-fold cross validations. The results of this study show that the highest accuracy is 92.98% at 5-fold cross validation using the RBF kernel with parameters C = 10 and gamma = 1. The performances of the three SVM kernels varied depending on the type of data used.