Abstract: To solve the problem that existing prediction models cannot extract the internal rules of the traffic flow in traffic big data and fail to make full use of spatiotemporal correlation characteristics to achieve high accurate prediction, a short-term traffic flow prediction model was proposed based on K-nearest neighbor (KNN)and long short term memory (LSTM). The KNN algorithm was used to select the stations that were related to the test station in the road network. The traffic flow data in selected stations were used as the training and testing datasets for the LSTM model. The proposed model was verified by real traffic data from the Transportation Research Data Lab in USA. Results show that the proposed method can better extract the spatiotemporal characteristics of traffic flow sequences, and the prediction accuracy can be improved by 12.28% on average compared with the existing prediction model, which can provide the necessary basis for traffic guidance and control.