An Extended Density-based Clustering Algorithm in Big Data

Today, data generation through smart equipment, including mobile phones, has faced a significant growth, and clustering is one of the most widely used knowledge discovery techniques in big data. Density-based clustering (DBSCAN) is one of the most efficient clustering algorithms in data mining, and despite having advantages, it also has problems, such as the difficulty in determining the input parameters, as well as not being able to detect clusters. with different densities. In the proposed algorithm of this article, it is inspired by the K-DBSCAN algorithm in grouping large data with the aim of reducing the clustering execution time.In addition, by using K-Means and H-DBSCAN algorithms, different densities of the data set were identified and an Eps radius was determined for each density, and then, the proposed density-based clustering algorithm was developed with parameters The matching is applied to the data, and in fact, the innovation of this article is the use of K Means clustering and the estimation of different densities in the DBSCAN clustering method. The proposed algorithm has been compared with the simple DBSCAN clustering algorithm and two developed K-DBSCAN and H-DBSCAN algorithms on four standard data sets: Image segmentation, Pendigit, Letters and Shuttle control. The results show that the proposed algorithm is superior to other algorithms when both time and accuracy are criteria in clustering.