The Comparison Between ARIMA and Long Short-term Memory for Highway Traffic Flow Prediction

抄録

With the complex traffic situation and worse traffic congestion, predicting traffic flow accurately is quite important. This thesis emphatically introduced long short-term memory model (LSTM) and time series models, especially the ARIMA model. Time series models had been widely used for traffic flow prediction many years ago. LSTM model now performs well in other prediction areas, but the traffic flow prediction doesn’t verify. In this thesis, we used highway traffic flow data to test model performance and compared them through several factors. The results showed that LSTM is well-suited in fluctuant traffic flow prediction and have high accuracy.