A study on DBSCAN Clustering algorithm issues and a survey on its improvements
Clustering is an important knowledge discovery technique in the database. Density-based clustering algorithms are one of the main methods for clustering in data mining. These algorithms have some special features including being independent from the shape of the clusters, highly understandable and ease of use. DBSCAN is a base algorithm for density-based clustering algorithms. DBSCAN is able to detect clusters with different sizes and shapes in huge amounts of data and is also resistant to noise. Despite its advantages, this algorithm has its own drawbacks such as the difficulty in determining appropriate values for input parameters, inability to detect clusters with different density and inability to detect appropriate clusters when they are too close.
Since 1996 that DBSCAN has been introduced, many different algorithms have been proposed as improvements of DBSCAN. In this paper, firstly the drawbacks of DBSCAN algorithm are discussed. Secondly, we review and discuss DBSCAN improvement algorithms in order to know the pros and cons of each algorithm and their success in improving DBSCAN algorithm. We also implemented some of these algorithms according to our studies and tested them according to the clustering evaluation criteria on standard data sets, so that we would to be able to judge the algorithms better.
Article Type:
Research/Original Article
Soft Computing Journal, Volume:6 Issue:1, 2017
2 - 37
روش‌های دسترسی به متن این مطلب
اشتراک شخصی
در سایت عضو شوید و هزینه اشتراک یک‌ساله سایت به مبلغ 300,000ريال را پرداخت کنید. همزمان با برقراری دوره اشتراک بسته دانلود 100 مطلب نیز برای شما فعال خواهد شد!
اشتراک سازمانی
به کتابخانه دانشگاه یا محل کار خود پیشنهاد کنید تا اشتراک سازمانی این پایگاه را برای دسترسی همه کاربران به متن مطالب خریداری نمایند!