Spatial Prediction of COVID-19 in China Based on Machine Learning Algorithms and Geographically Weighted Regression

Table 1

Summary of spatial independent variables.

Variables	Description	Data source	Format
Rh^1,∗	Relative humidity	ERA5	Hourly
T2m^1,	Temperature at 2 m height	ERA5	Hourly
MoveIn	City Migration index	Baidu Migration Map	Daily
MoveOut	City Emigration index	Baidu Migration Map	Daily
Travel	Intracity travel intensity	Baidu Migration Map	Daily; Daily
WP^2,	Traffic flow from Wuhan to other cities	Baidu Migration Map	Daily
WD^3	Geographic distance from each city to Wuhan	—	Yearly
GDP	GDP per city	NBSC	Yearly
People	Resident population per city	NBSC	Yearly

¹Calculation of daily average weather variables for each city. ²WP is constructed by multiplying MoveOut of Wuhan with percentage that a destination city receives from Wuhan for each Chinese city. For Wuhan, we set the . ³WD is constructed by calculating the geographic distance from each city to Wuhan under UTM ZONE 49N projection. For Wuhan, we set . Calculation of the average of independent variables for each city from January 17, 2020, to March 1, 2020.