Applying machine learning techniques in survival analysis to the private pension system in Turkey


ŞİMŞEK G., KARASOY D.

Communications in Statistics - Theory and Methods, 2023 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Publication Date: 2023
  • Doi Number: 10.1080/03610926.2023.2230329
  • Journal Name: Communications in Statistics - Theory and Methods
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Business Source Elite, Business Source Premier, CAB Abstracts, Compendex, Veterinary Science Database, zbMATH, Civil Engineering Abstracts
  • Keywords: accelerated failure time, Cox regression, Individual pension plan, machine learning, survival analysis
  • Hacettepe University Affiliated: Yes

Abstract

Problems such as the disruption of the income-expenditure balance and the decrease in active-passive ratio, which emerged at the end of the 1990s in Turkey, brought the need for reforms in the social security system. As a result of these reform efforts, a private pension system, complementary to the existing social security system, was put into practice. To our knowledge, no study has examined the private pension system using the Cox regression model, accelerated failure time models, and machine learning methods together under survival analysis. In this work, we show that machine learning methods provide non parametric alternatives to traditional survival models such as Cox regression. In addition to the statistics obtained, other important results are that socio-economic problems such as gender inequality, education inequality and income inequality can also be seen in private pension systems.