LSIS: Large scale instance selection algorithm for big data

Recently enormous volumes of data are generated in Information Systems, and data mining area is facing new challenges of transforming this “big data” into useful knowledge. To get from “big data” a manageable volume, we propose a large scale instance selection for reducing the initial dataset, leading to a reduction of both time taken and the computational resources that are necessary for performing the learning process, and improving the accuracy of classifier model. Our experimental results demonstrated that the proposed algorithms could scale well and efficiently process large datasets by selecting relevant instances for classification problem. The experimental results show also the contribution of the instance selection on the classification accuracy.