Prediction of Contribution Increase Status in the Private Pension System Using Machine Learning Methods

This study aims to predict whether participants in the Private Pension System (PPS) will increase their contribution amounts in the following year. The PPS plays an important role in enhancing long-term savings and generating sustainable economic resources. Participants’ decisions to increase their contributions directly affect both their individual savings and the overall fund size of the system. The dataset used in this study was obtained from a data science competition published on Kaggle in 2020 by an insurance company. The dataset includes participants’ demographic characteristics, payment behaviors, and financial indicators. The problem is formulated as a binary classification task to predict contribution increase behavior. In the analysis process, gradient boosting-based machine learning algorithms, namely XGBoost and LightGBM, were applied and their performances were compared. The findings indicate that contribution increase behavior is associated with certain behavioral and financial variables. In particular, monthly payments, average contribution amount, customer segment, and investment characteristics were found to be key determinants of model performance. In conclusion, the results demonstrate that machine learning methods can be effectively utilized in predicting participant behavior and supporting decision-making processes in the PPS.