Automatic clustering of big datasets using a swarm intelligence method

چکیده مقاله

Mining and discovering knowledge from big datasets have become a new interesting field of research among data scientists. In fact, extracting hidden patterns in big datasets using traditional data mining algorithms in a reasonable period of time and with an acceptable accuracy is impossible due to high volume of data and their complexity. Generally, the term big data is referred to massive datasets with huge number of high dimensional samples which makes them very hard to be analyzed by conventional data mining techniques. So designing new and effective algorithms for analyzing big datasets is necessary. Clustering, which is the process of dividing the data points into different groups based on their similarities and dissimilarities, is one of the most important data mining and big data mining methods. K-means, which is one of the most popular clustering algorithms and has been widely used in several researches, suffers from some drawbacks such as: its tendency to converge to a local optimum point, the quality of its final results depends on the initial centroids generated randomly and its inability in finding the number of clusters. In this paper a new automatic big data clustering method, based on a swarm intelligence algorithm, is introduced which has a great ability in finding the number of clusters and escaping from local optimum point. The proposed method is tested on 13 synthetics and 2 real big mobility datasets. Final results demonstrate its power in big data clustering.