Identifying different types of urban land use dynamics using Point-of-interest (POI) and Random Forest algorithm: The case of Huizhou, China

Abstract

The significant economic development witnessed in China in recent decades has been accompanied by the increasing expansion of urban areas. Although a growing literature has analyzed the characteristics and driving forces of urban land expansion, less attention has been paid to examining the different expansion determinants driving fine-scale urban land use (residential land, administration and public services land, commercial land, and industrial land) change. This paper aims to identify the differences of multi-mechanisms driving fine-scale urban land use expansion based on big data and machine learning, in the Huizhou downtown area in 2000–2015. The Random Forest (RF) algorithm is used to identify the natural, transportation, location, social, and POI factors driving land expansion by considering different urban land-use categories. Our RF estimations showed that enormous differences existed between various urban land-use types in terms of the role they played in this expansion and their relation to potential determinants, during the different urban development stages studied. Transportation, location, and the distribution of actual land use were found to exert a greater influence on urban land expansion than other factors. All the findings above provide detailed spatiotemporal knowledge and targeted information that can aid in understanding fine-scale urban land use dynamics. In this way, sound planning strategies for different fine-scale land uses can be formulated more scientifically. The strength of association between these factors and urban land expansion differed greatly depending on the different land-use types involved as well as the urban development stage that it occurred within. These results cast a new light on the importance of investigating the potential driving forces in the expansion of different urban land-use types.