Urban Function Identification Based on POI and Taxi Trajectory Data

Abstract

In order to effectively implement urban planning decision-making behavior, urban function analysis is necessary. Based on POI data which can reflect the static characteristics of regional functions and taxi spatio-temporal trajectory data which can reflect the urban functional activity, this study proposes a non-negative matrix decomposition (KNMF) model with kernel function, and uses this model to identify urban functions. In the framework of KNMF algorithm, the urban functional areas are identified quantitatively, and the sensitivity of the parameters in the model is analyzed. Finally, the recognition results of the proposed method are compared with those of the DMR topic discovery model, which proves the effectiveness of our algorithm.