[Application of Support Vector Machine Regression in Ozone Forecasting].

Support vector machine regression (SVMr) was proposed to forecast hourly ozone (O) concentrations, daily maximum O concentrations, and maximum 8 h moving average O concentrations (O 8 h) by employing the observations of meteorological variables and O and its precursors during the high O periods from May 20 to August 15, 2016 at an industrial area in Nanjing. The squared correlation coefficient () of the hourly O concentrations forecast was 0.84. The mean absolute error (MAE) and mean absolute percentage error (MAPE) were 3.44×10 and 24.48, respectively. The key factors for the hourly O forecast were the O pre-concentrations, amount of ultraviolet radiation B (UVB), and the NO concentration. The main factors for the O daily maximum forecast were the NO concentrations at 07:00 and the UVB level. Temperature and UVB played an important role in predicting O 8 h. In general, taking precursors into account could increase the accuracy of O prediction by 10%-28%. For O concentration forecasting, SVMr gave significantly better predictions than multiple linear regression methods.