A neural network is trained to predict the air quality in Beijing based on the concentration of certain particles, the temperature, precipitation, wind speed, etc.
The dataset used contains hourly air pollutants data from 12 nationally controlled air-quality monitoring sites. The air-quality data are from the Beijing Municipal Environmental Monitoring Center. The meteorological data in each air-quality site are matched with the nearest weather station from the China Meteorological Administration. The time period is from March 1st, 2013 to February 28th, 2017.
There are a total of 15 features used to predict the PM2.5 concentration
- year: year of data Input Variable
- month: month of data Input Variable
- day: day of data Input Variable
- hour: hour of data in this row Input Variable
- SO2: SO2 concentration (ug/m3 ) Input Variable
- NO2: NO2 concentration (ug/m3 ) Input Variable
- CO: CO concentration (ug/m3 ) Input Variable
- O3: O3 concentration (ug/m3 ) Input Variable
- TEMP: temperature (degree Celsius)
- PRES: pressure (hP a)
- DEWP: dew point temperature (degree Celsius)
- RAIN: precipitation (mm)
- wd: wind direction
- WSPM: wind speed (m/s)
- station: id of the air-quality monitoring site
- PM2.5: PM2.5 concentration (ug/m3 ) Output Variable