Visual analysis of spatio-temporal point patterns –A case study of locations of grocery shops in the Helsinki metropolitan area

Location is of vital importance to a shop. Since 1990, there has been a significant change in the makeup of Helsinki metropolitan area shop market: the number of grocery shops in the Helsinki metropolitan area remarkably decreased, while the number of hypermarkets steadily increased. In order to understand and explain the phenomena, both grocery shops and hypermarkets in the Helsinki metropolitan area which existed from 1990 to 2005 were mapped and analyzed. The study aimed to figure out patterns. The aim of the study was firstly to find out location relationships between grocery shops and hypermarkets and analyze patterns among grocery shops and, secondly, to evaluate different visualization methods basing on their usefulness in the analysis. To achieve the research objectives, different visual methods and analysis methods were applied to the case study dataset. The visual methods employed include single static maps, a time series of static maps, 3D visualization and animated maps. The analysis methods employed in this study included kernel density analysis, normalization, spatial join and Thiessen polygon analysis. The dataset used in the case study records locations and existences of both hypermarkets and grocery shops in the Helsinki metropolitan area from 1990 to 2005, but was simulated from the real shops dataset. Because of copyright and commercial protection, the real shop dataset was not allowed to be used in this thesis. However, this study documents the analysis processes and methods which could be used in and provide guidelines for the real dataset and also further research. The findings of this study indicate that in the Helsinki metropolitan area, the locations of grocery shops have been influenced by the locations of hypermarkets, the density of grocery shops and the density of population. But the development of grocery shops is a complex system more influencing factors need to be considered in the future studies. In terms of the visualization methods, in this specific case study, single static maps and time series of static maps were the most useful visual methods and normalization was an important part of the analysis method. Compared to the static visual method, 3D visualization and animated maps were more attractive and allowed more interaction with users but are not recommended for further analysis because they are not as effective as static visual methods for analyzing patterns and relationships of the sparely spread point dataset, like the locations of grocery shops