Clustering of Domestic Locations in Layers for the Purpose of Breakout Prevention of Diseases

Abstract

The aim of this research is to help support Provincial Public Health Offices (PPHOs) by clustering geographical coordinates in terms of latitudes and longitudes and comparing each cluster’s density in order to prevent mosquito-born diseases. In this research we propose using the hierarchically spatial clustering approach with two layers of clustering. We cannot use pre-existing clustering algorithms because they do not fit our purpose due to the naturally breeding area of common house mosquitoes. Thus, a modified method with an additional strict constraint of k-means and k-nearest neighbors were applied to meet this goal. Additional constraint to k-means clustering in the first layer will cause some of the coordinates to un-cluster. Finally, all coordinates must be clustered again to a second layer. However, the constraint of mosquito’s breeding area must be retained thus some falsely alarmed coordinates must be filtered out.