A Study on Spatial Distribution and Changes of Taiwan's Population by a Radiation Model

Abstract

In urban and regional policy (making) human movement is one of the key factors to consider Gravity Model is the most used approach to predict mobility patterns of people (Over the past few decades ) The model has been improved by previous studies and has achieved a major accomplishment in population movement prediction However Gravity Model still has its limitations Radiation Model which is applied to predict mobility fluxes between two regions has been developed in recent years and has been expected to become a new approach to sketch out the human migration patterns This study has two main objectives The first is to apply the Radiation Model to predict human migration in Taiwan under two different spatial scales The analysis under large spatial scale predicts the migration across cities and counties and the study with smaller scale demonstrates the patterns of movement with short distance in a region The second is to investigate whether the calibration based on historical data affects the ability of prediction of two human movement models The comparison between Radiation Model and Gravity Model has been made in the study Population migration statistics between cities counties and municipalities in Taiwan has been used to conduct the empirical analysis Result shows that the prediction of Radiation Model is not always better than Gravity Model under different statistical verifications On the other hand the calibration of historical data can drastically improve the outcomes of both models The reason could be the limitation of the data The data coverage is limited to 13 counties and cities and 6 municipalities and parts of districts in Tainan City A more detailed data covers the migration statistics between all administrative areas in Taiwan would be useful to thoroughly examine the feasibility of applying Radiation Model in Taiwan Also big data such as mobile probes can provide more precise data on population fluxes and can solve small sample size problem in previous study

Date of Award	2020
Original language	English
Supervisor	Oliver Feng-Yeu Shyr (Supervisor)