Urban land use function prediction method based on RF and cellular automaton model

Abstract To accurately grasp the dynamic changes of urban land use and solve the difficulties and challenges in predicting urban land use functions at present, this study integrates interest point data and open street map data through kernel density estimation technology. Moreover, the study also integrates random forest algorithm and cellular automaton model, and finally proposes a new urban land use function prediction method based on random forest algorithm and cellular automaton model. The experiment results show that the comprehensive precision and Kappa coefficient calculated by the research method reach 81.88% and 0.71, separately, verifying the validity of the way. The prediction results of this method indicate that the number of squares required for road and transportation, industrial land, public services, residential land, green squares, and commercial service land in Hulunbuir City in 2030 is expected to reach 2000, 3889, 2591, 9280, 2696, and 8988, respectively. This provides a scientific basis for future urban planning. To sum up, the method raised by the study has high applicability and accuracy in predicting the distribution pattern of urban functional regions, and has important instructing significance for urban planning, optimal assignment of land resources, and continuous expanding of urbanization.