Air Quality (AQI) Monitoring and Prediction based on Trajectory Data
The issue of air quality is a hot topic in the area of environmental protection in recent years. Due to the high cost of air quality inspection station, the number of stations is low and limited and the distribution is not uniform throughout the city. At the same time, the Air Quality Index (AQI) is not completely uniform and continuous in the geographical distribution. Overall, there have been large differences between the actual air quality conditions and the number of nearby air quality inspection station in some parts of the city. Based on this problem, because there is a certain similarity between the components of vehicle emissions and that of air quality indication, I introduce taxi trajectory data to forecast AQI which based on the air quality index of nine air quality stations in New York City from March to April 2014, combined with meteorology data and point of interest(POI) data. In this paper, the gridding of cities is firstly carried out. The regression analysis shows that the trajectory data is related to AQI as well as its index items. After using panel data regression model to extract individual differences and time differences, multiple classification models are used to predict AQI and compare the actual AQI data to test effects, including model integration. Finally, the model achieves fine-grained (on the level of 80% accuracy) prediction of air quality index AQI for every part of the city.
Trajectory data mining, Air Quality Index, Classification model, Grid Partition
Prof. Shengsheng Xiao