Forecasting air quality index in Thailand using ensemble method

Abstract

Air pollution is one of the most important problems that needs to be urgently solved around the world. Inevitably, Thailand has had to fight against it as well, particularly in the Northern Thailand. This region also has faced high air contamination for several years. In this thesis, the proposed model based on ensemble method was presented to predict the extent of air quality index (AQI) in the region from the majority vote of outputs from three classification algorithms, namely, support vector machine, random forest, and k-nearest neighbors. This proposed method made a comparison between the voted classification accuracy and the accuracies of the individual classification models. The model took advantage of seven datasets from monitoring stations in four provinces in the Northern Thailand. Eventually, the proposed ensemble model produced, on average, the accuracy rate of 99.68% - 99.84% greater than most of the accuracies of the other comparative models.