Traffic Speed Prediction Method Based on Spatiotemporal Sampling and LSTM

Abstract

Accurate prediction of traffic speed is crucial for traffic management and planning. In order to solve the problems of low prediction efficiency of previous traffic speed prediction models and easy neglect of spatiotemporal characteristics, a traffic speed prediction method based on spatiotemporal sampling and LSTM model is proposed based on the 24-hour driving dataset of 4,000 taxis in Shanghai, and draw speed heat maps in different regions at different times to visualize the spatiotemporal characteristics of traffic speed. Experimental results show that the model has high prediction accuracy and good expansion potential.