Improving Logistics of the Public Services in Smart Cities Using a Novel Clustering Method

Abstract

Smart City public services need detailed and relevant public information to increase their efficiency. To have relevant information, collecting and processing the data about its previous uses are crucial. Clustering is one of the most powerful, yet computationally demanding, tools that can be used to process such information. Since public services data are vast, but usually not accurate, the objects clustered are considered as uncertain. In this paper, we propose a novel clustering method for uncertain objects called Improved Bisector Pruning (IBP), which uses bisectors to reduce the number of computations. We combine IBP with a modified segmentation of a data set area (SDSA) method that enables the parallelization of the clustering process. In the experiments, we show that IBP-SDSA is superior in performance to the most used clustering method UK-means combined with Voronoi or MinMax pruning, regardless of the problem size. We applied IBP-SDSA on clustering the public services data in the city of Osijek and show that the acquired data can be used to improve public services logistics.