Incremental clustering algorithm based on representative points and covariance for large data

As the dynamic data increases, more space is needed to store the data. However, most traditional clustering methods are time-consuming and only suitable for static data. For this problem, incremental clustering methods are increasingly used in dynamic data. The study proposes an incremental clustering algorithm based on representative points and covariance for large data (IDPC_RC). Firstly, the representative points were selected in the initial data. Then, the similarity between new data points and representative points was calculated to find the pre-allocated cluster. Finally, the covariance determinant was used to measure the degree of local imbalance for pre-allocated clusters after new data is added, and the cluster numbers were adjusted adaptively. The performance of the proposed scheme was tested on five benchmark datasets and real consumption data. The experimental results show the scheme achieves excellent clustering performance and low time consumption on all datasets, which is useful for incremental clustering tasks. Copyright © 2023 Inderscience Enterprises Ltd.