M. Indulska and M.E. Orlowska (Australia)
clustering, facility location, spatial, GIS
Clustering is used to group elements of a set based on some similarity measure of those elements. It can be useful for the facility location group of problems, where k locations for facility placement are required to be found in order to minimize access distance for most customers. Such facility location problems require a solution which takes into account a region with its natural topography, residential objects (usually modelled as points), the 'capacity' of those objects access paths (such as roads and footpaths) as well as the existence of obstacles such as lakes or rivers. The main purpose of this short paper is to revisit existing solutions to spatial clustering with obstacles. We illustrate that the current consideration of obstacles in clustering [1,2] is not realistic and show that predefined access paths should be taken into account when solving facility location problems. Additionally we present a discussion of issues that need to be considered when clustering in spatial environments. We also argue that the problem of clustering with obstacles, as shown in [1,2] can be reduced to its version without obstacles and thus can be solved with existing clustering algorithms with no need for any extension.
Important Links:
Go Back
