The Improved Leslie Model for population Forecasting

Abstract

Based on China’s population data from 1953 to 2020, the Leslie model combines the fertility rate of women of childbearing age by region and age, the sex ratio of the birth population, the mortality rate, the migration rate between urban and rural areas by age, the curve fitting migration function, and the application of ARIMA to predict mortality rates to construct a discrete population dynamics system in order to predict China’s future population development trajectory. The improved Leslie, Leslie, BP and Malthus models were compared in terms of error rates. The improved Leslie model was more stable than the rest of the models and had an average error rate of 0.09%, with good model generalization ability. The results show that the improved Leslie model predicts that the total population will slowly increase under the national regulation policy, and will reach a peak by around 2045 and then decline.