Automatic Traffic Anomaly Detection on the Road Network with Spatial-Temporal Graph Neural Network Representation Learning

Table 1

Results of comparison with previous work.

Methodology	Research contents	Multiple variables		Data relationship
		Yes	No	Temporal	Spatial	Temporal-spatial
Statistics-based method	The autoregressive integrated moving average model (ARIMA) [6]		√	√
	The local outlier factor (LOF) [19]	√				√
Machine learning-based method	k-nearest neighbor (KNN) [8]	√			√
	Long short-term memory (LSTM) (Zhao et al.,2017)	√		√
	Graph convolutional network (GCN) [20]	√			√
	This study	√				√