In view of the bus load forecast features such as large amount and wide range

susceptibility to weather changes

plus a very loose geographical correspondence between traditional weather forecast and the bus load

a new bus load forecasting method based on numerical weather prediction and load classification is proposed. By learning about the respective location in the administrative area of bus load

such as counties

townships or municipal districts

the corresponding high precision numerical weather information is extracted. Moreover

by studying the bus load and meteorological factors related features

the bus load forecasting model is built based on load classification. The method has successfully been applied in a provincial bus load forecast.