利用LSTM-Transformer的城市小区入住率预测OA

In the current stock era,the housing demand of urban residents tends to be diversified and personalized.The occupancy rate prediction of urban neighborhoods is not only important for real estate developers,governments and urban planners,but also can reveal the economic vitali-ty,population migration trend and social development potential of urban areas.In this paper,based on mobile phone signaling data combined with multi-scale socio-economic data,taking Jinan city,Shandong Province,China for example,we proposed an urban occupancy prediction model based on LSTM-Transformer,which was expected to fill in the gaps in the existing research,and explored the spatial distribution character-istics of occupancy rate in the city.The experimental result shows that the model is better than other traditional models in terms of prediction accuracy and has high reliability.The residential occupancy rate of Jinan City shows a decreasing trend from the central city to the outside,and the occupancy rates of the central and eastern areas are higher than that of the southwest.In terms of spatial distribution,there are high-value and low-value gathering areas in Jinan City,showing a change from"high-high"to"low-low"gathering mode from the downtown to the periphery.The prediction framework constructed in this study not only provides data support for precision real estate marketing in the stock era,but also pro-vides scientific basis and decision-making reference for the government to optimize urban public resource allocation,identify urban shrinkage risks,and promote fine-grained management of smart cities,which has significant social application value.