An Efficient Enhanced Algorithm for Clustering Large Datasets

This is the authors' abstract. We don't add key points for this paper.

Clustering in data mining is a powerful tool for gaining knowledge. There is tremendous knowledge in the field of clustering as it has recently appeared in several different applications. The density-based method is a successful technique for large datasets (big data). Algorithms such as the well-known DBSCAN algorithm can find clusters of random shapes defined as high- and low-density regions. This paper presents an enhanced algorithm (called eDBSCAN) that works on large data sets as it can handle large sets of data with high accuracy. This is to obtain the set number of groups of clusters (K) on the experimental approach in the traditional DBSCAN algorithm. We used the classified data set to avoid the main problem of the partitioned technique that clarifies the number of K, i.e., clusters. An experimental study was conducted to prove the effectiveness of the proposed algorithm in finding the value of Epsilon parameters (Eps), where the algorithm was applied to a synthetic data set and it was proven that the eDBSCAN algorithm achieved scalability with large data sets and a high degree of purity.