A Study on Chinese Residents' Willingness to Delay Retirement Based on Machine Learning Algorithm

Abstract

At present, the problem of population aging caused by the increase of average life expectancy continues to worsen, and thus leads to a series of problems. The implementation of a delayed retirement policy in China will affect everyone. In this paper, the data of China Health and Pension tracking survey in 2018 are used. Firstly, the optimal parameter combination of deep neural network, RF, GBDT and XGBoost are obtained by grid search algorithm to conduct single model modeling. On the basis of the above four single models, linear effects are introduced by logistic regression. By using the idea of model fusion, a new prediction model is obtained by weighted fusion learning of the above five models. The important factors affecting the public's retirement intention are found and the high precision prediction of retirement intention classification is carried out.