Rédiger un PREreview

The accurate identification of meaningful patterns in high-dimensional and noisy datasets remains a fundamental challenge in intelligent data analysis, particularly within the domain of smart city analytics. Traditional clustering algorithms such as DBSCAN offer robustness to noise and the ability to detect clusters of arbitrary shapes. However, they suffer from critical limitations, including sensitivity to parameter selection and poor performance in handling overlapping or ambiguous data regions. To overcome these issues, this paper presents a novel hybrid clustering framework that synergistically combines fuzzy logic and Particle Swarm Optimization (PSO) with Density-Based Spatial Clustering of Applications with Noise (DBSCAN). The proposed method begins with Z-score normalization for data standardization, followed by the application of PSO to automatically optimize key DBSCAN parameters, namely Eps and MinPts, across a predefined range. A fuzzy extension of DBSCAN is then employed to enable soft clustering, which better accommodates data uncertainty and overlapping class boundaries. Experimental evaluations on urban analytics datasets from Addis Ababa demonstrate that the proposed approach achieves improved clustering quality, as evidenced by enhanced silhouette scores and intra-cluster cohesion, in comparison to traditional DBSCAN and its variants. This work contributes a flexible and intelligent clustering technique well-suited for real-world smart city applications where data ambiguity and parameter sensitivity are prevalent.