A SEMI PARAMETRIC AUTOREGRESSION MODEL FOR PREDICTING AIR QUALITY

Air quality has become a hot issue of general concern, and air quality analysis and modeling have received extensive attention. There are many factors affecting air quality, mainly including natural factors and human factors. Air quality and gas pollution are mainly based on ordinary ARMA models. Based on the quasi linear variable coefficient spatial autoregressive model and estimation method, the air quality in Harbin is analyzed. The natural factors such as temperature, humidity, pressure, wind speed, and the non natural factors such as heating are selected as the influence variables of the model. The model is established with nitrogen dioxide, ozone, carbon monoxide, two types of respirable particulate matter (PM10 and PM2.5), and sulfur dioxide as the response variables. The model results show that some pollutants have obvious relations with natural factors and human factors, while others have no obvious relations.