#$%^&*AU2020100707A420200611.pdf#####Abstract This paper presented a specific model based on linear regression analysis to predict the haze condition. Actual data were adopted as samples. Firstly, particular sensors were used to collect the data describing the haze condition from April to October in Beijing. The data include (1) atmospheric concentration data, such as the amount of PM1.0 and PM2.5; (2) meteorological data, such as temperature and humidity. Average values of these data were divided into train set and test set proportionally. During learning stage, data in train set were used to build linear regression model, which is used to forecast the future tendency of several elements in haze. Data in test set acts as samples for reference to test the accuracy of the model built previously. This model can help predict the haze condition in Beijing in a day or a week. It can also fill up the lost data value when the sensors failed to collect it. 1
展开▼
