JIITA, vol.3 no.3, pp.288-293, 2019
DOI: SOON
A study of particulate matter clustering for PM10 distribution prediction
Seonghee Min 1), Yoosoo Oh 1,*)
1) School of Computer & Communication Engineering, Daegu University, Gyeongsan-si, Republic of Korea
Abstract: In this paper, we propose the clustering algorithm to estimate particulate matter distribution. The algorithm divides the area of the center that the fine dust distribution using K-means clustering. And then it finds the coordinates of the optimal point according to the distribution of the fine dust values. We use the particulate matter data in the AirKorea site provided by the Korea Environment Corporation. First, we downloaded particulate matter data of the country to generate feature datasets. Second, we convert observatory address into the latitude and longitude of WGS84 coordinates. Lastly, we performed the K-means clustering algorithm to cluster feature datasets. Feature datasets are latitude, longitude of observatory and PM10 values. In this paper, we conducted an experiment on the K values to better represent the cluster. We performed clustering by changing K values from 10 to 23. Then we generated 16 labels divided into 16 cities in Korea and compared them to the clustering result. By visualizing them on the actual map, we confirmed whether the clusters of each city were evenly bound. Moreover, we figure out the center of the cluster to find the observatory location that can represent particulate matter distribution.
Keyword: Particulate matter; Clustering; PM10; Fine dust distribution clustering; K-means clustering
A study of particulate matter clustering for PM10 distribution prediction