Big-Data Clustering with Genetic Algorithm

The data emerging from Internet of Things (IoT) usually exhibits a wide variety and is big in terms of quantity, i.e., Big-Data. As a preprocessing step, clustering is a regular task in such systems and normally is done using evolutionary algorithms. In this paper, we propose a new automatic clustering algorithm based on Genetic Algorithm (GA), in which, it is NOT mandatory to know the number of clusters. The proposed algorithm uses a very short chromosome encoding and proposes relevant crossover and mutation operators that lead to a very good clustering performance. Our algorithm, uses an unsupervised learning paradigm to classify the data points into clusters. To demonstrate the performance of the proposed algorithm, it is evaluated with balanced/unbalanced real-world data containing 13-tuple data vectors, and also with a 1.000.000-sample artificially generated random data set. At either cases, our algorithm outperforms the other algorithms.