[Retracted] A Multiscale and High-Precision LSTM-GASVR Short-Term Traffic Flow Prediction Model

Table 8

Analysis of short-time traffic flow prediction algorithm for different prediction algorithms.

Algorithm	Summary
LSTM	The prediction effect is good and timeliness is general, but there are big differences for different models trained
GRU	Simpler than the LSTM structure, but the traffic flow prediction at the turning point is less effective
CNN	It is suitable for predicting stable traffic flow, and the prediction effect is not good when the volatility is large
SAE	SAE model has large prediction error when traffic flow data is small
ARIMA	ARIMA model prediction error is general and timeliness is poor
GASVR	The prediction effect is good and timeliness is better, but there is a certain shift phenomenon
LSTM-GASVR	The timeliness of the model is normal, and a variety of time interval data prediction effect is better and more stable