Evaluation of urban residential land use efficiency with a neural network from the perspective of service facility capacity

ABSTRACT We demonstrate a method to evaluate the urban residential land use efficiency from the perspective of service facility capacity by using a neural network. Publicly available data were used to obtain the physical and socioeconomic information. Combining the actual carrying capacity calculated from the residential population and the theoretical carrying capacity calculated from the service facility density, we trained a back-propagation neural network to evaluate the efficiency of residential land use. The results show that the degree of residential land use can be reflected through the carrying capability of service facilities. Beijing’s residential land efficiency presents a spatial distribution form that declines from the centre to the edge. We also found a phenomenon that the efficiency of the residential land at the junction of administrative districts is relatively low. Our research demonstrates how multi-source publicly available data and neural network algorithms can be applied to solve complicated social issues.