Performance Analysis Of a Parallel Implementation of the City Hotspot Detector Algorithm

Geo-referenced data are a valuable source to detect hotspots in urban environments. These hotspots represent urban events such as crimes, disease outbreaks, and road traffic accidents that occur at higher densities compared to other areas, supporting informed decision-making for urban management. Since urban areas exhibit heterogeneous densities, multi-density clustering methods are more suitable for discovering urban hotspots than classic density-based methods. Furthermore, due to the high volume of these data, clustering algorithms can benefit from a parallel implementation, enabling scalable computations that improve both execution time and computational efficiency. This paper analyzes the improvements in terms of execution time, speed-up, and efficiency of a parallel implementation of the multi-density clustering CHD (City Hotspot Detector) algorithm.