Air Quality Prediction Model Based on Spatiotemporal Data Analysis and Metalearning

Table 1

Symbols and explanations.

Symbol	Meaning
	Set of monitoring stations in the target city
	Length of the historical data
t	Current time
	Length of the forecast window
	Concentration of target pollutant monitored by station s at time (past)
	Historcical data vector of station
	Historcical data vectors of all stations
	Historical weather of the target city at time
	Historical weather dataset
	Weather forecast of the target city at time
	Weather forecast dataset
	Set of training dataset from source cities
	Training dataset from the target city
	Concentration of target pollutant monitored by station s at time (future)
	Concentration of target pollutant at station s in the future
	Prediction model with parameter
	Loss of model on training batch
	Predicted value of .
	Influence radius of monitoring stations
	Directed graph built on all the monitoring stations
	Output vector of the LSTM unit on station at time
	Cell state vector of the LSTM unit on station at time
	Output vector of GAT on station at time
	GAT’s similarity score between node and
	Weight of the edge in calculated by GAT