KIDBSCAN: A new efficient data clustering algorithm

Spatial data clustering plays an important role in numerous fields. Data clustering algorithms have been developed in recent years. K-means is fast, easily implemented and finds most local optima. IDBSCAN is more efficient than DBSCAN. IDBSCAN can also find arbitrary shapes and detect noisy points for data clustering. This investigation presents a new technique based on the concept of IDBSCAN, in which K-means is used to find the high-density center points and then IDBSCAN is used to expand clusters from these high-density center points. IDBSCAN has a lower execution time because it reduces the execution time by selecting representative points in seeds. The simulation indicates that the proposed KIDBSCAN yields more accurate clustering results. Additionally, this new approach reduces the I/O cost. KIDBSCAN outperforms DBSCAN and IDBSCAN.