Exploiting Regressive Model for Population Prediction in China

Abstract

The demographics of China reveal a huge population, which amounted to around 1.4 billion people in 2022. However, evidence and authorities demonstrate that the Chinese population is about to shrink for the first time and the country’s total fertility rate keeps decreasing. It introduces great uncertainty to the future development of China. An accurate population prediction is important to estimate the economy, make reasonable policies, and stabilize society. In this paper, three machine learning models, including the polynomial regression, logistic growth model and Autoregressive Integrated Moving Average (ARIMA), are used to forecast the population in China. The effectiveness of different models on population forecasting is compared and analyzed. The results show that ARIMA performs the best, which is about a 0.34% error rate validated on previous population data. The prediction results demonstrate that the population in China will experience a brief rise and then enter negative growth. To embrace the population decline, the government should get prepared for the aging society and propose reasonable policies to increase fertility.