Adaptive Dynamic Clustering for Streaming Data using Genetic Algorithm

Dynamic clustering is a critical task for applications involving streaming data, such as customer behavior analysis and real-time pattern recognition. Traditional clustering algorithms like KMeans struggle to adapt to evolving data distributions, especially in non-convex or noisy environments. This paper presents a Genetic Algorithm (GA)-based dynamic clustering framework designed to handle such challenges effectively. Our method encodes cluster centers as chromosomes and employs a fitness function that minimizes the Sum of Squared Errors (SSE) while penalizing unassigned clusters. We evaluate the proposed method on both synthetic datasets (e.g., moons and Gaussian blobs) and a real-world customer dataset. Experimental results demonstrate that our GA-based approach outperforms KMeans in adaptability and robustness across time steps, with visual evidence and fitness score comparisons. The proposed system highlights the potential of evolutionary algorithms in adaptive learning scenarios, particularly in dynamic data stream environments.