Early Detection for Determinant Factor of National Health Insurance Membership for Workers in Central Java Using C4.5 Algorithm

¹ Faculty of Health Science, Universitas Dian Nuswantoro, Semarang, Indonesia
² Faculty of Economics and Business, Universitas Dian Nuswantoro, Semarang, Indonesia
³ Faculty of Economics and Business, Universitas Indonesia, Depok, Indonesia

^* Corresponding author: aprianti@dsn.dinus.ac.id

Abstract

Central Java Province has a membership coverage rate below national coverage. Several data sources up to October 2023 show that NHI participation coverage in Central Java is 93.13%, with 72.72% active participants. The highest type of participation is Contribution Assistance Recipients. A survey of 451 tobacco workers in Central Java found that 55.2% did not have health insurance. This research aims to apply the C4.5 algorithm for the early detection of determinant factors of national health insurance membership for workers in Central Java. Research was done in Central Java districts: Semarang City, Semarang Regency, Jepara Regency, and Kendal Regency. These four districts were chosen because they have the largest workforce in Central Java Province. The total sample was 400, with 100 respondents taken from each city/district. The sampling technique was cluster sampling; two sub-districts were taken from each district—data analysis using the Data Mining 4.5 algorithm. The results show that employment status is the most influential factor in determining someone to become an NHI participant, where fishermen are the group of workers most predicted not to become NHI participants with a gain value of 0.3887. Measurement using RapidMiner software proves that the C4.5 algorithm has an accuracy of 88.93%. At the same time, the AUC curve has a value of 0.941, which, according to Gorunescu, is included in the Superior Classification. The program is expected to prioritize fishermen groups as members of the NHI program.