Rough-graph-based hotspot detection of polygon vector data

Spatial polygon data represents the area of some events such as disease cases, crime, health care facilities, earthquakes, and fires. Finding the hotspot is crucial in exploratory data analysis. Although finding the spatially significant cluster is still challenging work. On this account, in this paper, we proposed a novel method based on the rough graph that finds the statistically significant hotspot. First, Global Moan's I index is calculated to find the presence of a hotspot in the data set. A positive value of Global Moran's I index shows the presence of a hotspot in the dataset. Then, the RGBHSD algorithm is used, which constructs a rough graph by considering each polygon as the node, and there is an edge between the two nodes if two polygons are neighbours of each other. Then boundary value analysis is done on the lower region of the rough graph, which considers some more boundary value polygon to be changed as the lower region. The polygons belonging to the lower region are considered the candidate hotspot. After detecting the candidate hotspot, a statistical significance test is done to find the significant hotspot. Finally, the RGBHSD algorithm is evaluated based on the evaluation metrics. We tested the algorithm on the socio-economic dataset of UP, India and Brexit dataset of UK. In the socioeconomic dataset the health facility provided in the villages is used to find the hotspot. In the Brexit dataset field related to the percent of the vote for the UK to be in the European union or not is taken. After the analysis, it is found that the hotspots generated are denser, and the time taken by the algorithm is less and the HPAI value is high than other literature methods. The result shows that the hotspots are scattered over the study region but clustered in some areas like west UP, east UP, etc. The hotspot offers health facilities in these virtuous areas and for Brexit data hotspot is clustered in the south region. This type of analysis is suitable for dealing with the pandemic, and to understand the pattern of any disaster drought, flood etc.