Using Non-negative Matrix Factorization to study the air pollution in central Taiwan regions

• Main Authors: Jui-Fang Chang, 張睿舫
• Other Authors: Yin-Tzer Shih
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi/login?o=dnclcdr&s=id=%22107NCHU5507028%22.&searchmode=basic

碩士 === 國立中興大學 === 應用數學系所 === 107 === In this thesis, we analyze the main sources of pollution for SO2, NO2, PM10 and PM2.5. We using non-negative matrix factorization (NMF) to perform dimension reduction on the data and using cophenetic correlation to determine the number of components. Then we visualize the results by adding wind direction and wind speed to determine whether the pollution is transboundary or domestic. Our results show that the pollution of PM2.5 and NO2 in central Taiwan area is mainly affected by domestic factors, and SO2 is mainly affected by transboundary factors. The domestic and transboundary pollution concentration ratio of PM10 is almost the same.