APPLYING DATA MINING TECHNIQUES IN URBAN COMPUTING FOR SMART CITIES

Abstract

To transform a city into a smart city it is important to focus on civic issues faced by the inhabitants. Civic complaints incorporate problems related to street condition, traffic, noise, water etc. Their analysis can contribute in proactive decisions to be taken by the city planners. Urban Computing is applied in many areas like transportation, environment, and security etc. but there is a need to explore more on urban planning from the perspective to analyze root cause of civic issues and reducing their concentration. In the present work, segregation of civic complaints based on different urban areas has been done and civic issues critical in an urban region are determined. For this purpose two approaches have been proposed namely Dynamic grid based clustering (DGCA) and clustering based on zip code approach (CZCA). A two phase clustering has been performed for both of the proposed approaches. The Phase 1 is different for both the approaches whereas the Phase 2 is similar. The purpose of Phase 1 is formation of spatial clusters. In DGCA, Phase 1 comprises of breaking the metropolitan area into grids representing spatial clusters. The granularity of the grid is determined by density of the civic complaints. In CZCA, Phase 1 comprises of dividing the civic complaints based on the zip codes of the region. The purpose of Phase 2 which is common for the two approaches is formation of sub-clusters based on complaint category over the spatial clusters obtained in Phase 1. These sub clusters are further analyzed to determine regions of city imitating similar complaint behaviour and finding the criticality of different complaint categories. For the purpose of experiment the real world dataset have been used for multiple metropolitan cities for USA and India. Experimental results have also been visualized to show better interpretation and compared with standard clustering algorithm and real world ground truth. The results are very promising and will help in planning strategies to improve inhabitant’s satisfaction rate and consequently improving their quality of life.