Bigdata-based investigation of heterogeneous migration and population distribution

Abstract

Abstract Human mobility is very important in understanding complex social and economic systems. With massive empirical datasets from the China Household Finance Survey and the National Statistics in the UK, we construct a migration probability matrix, and analyze the heterogeneous migration patterns. We then develop a random walk model to dynamically simulate the population distribution. In the stationary state, the resulting population distribution is in good agreement with the real statistical data. For comparison, simulations with an optimized gravity model and other datasets such as the census data in China are also performed. Further, the model simulation is applied to predict the demographic trend with different education levels. Our method could be generally extended to other real communities and internet worlds.