Abstract. Solving the problem of urban traffic congestion has always been an important issue in urban development. By predicting the congested areas in the city, it can be dealt with more timely and alleviate traffic difficulties. This paper presents a traffic congestion prediction model based on local correlation of urban cells. Reducing the granularity of the analysis to the cell can effectively improve the efficiency of the area, and the local correlation takes into account the effect of regional and interregional traffic loading. The experimental results show that the accuracy and recall rate of the proposed model are good.