Online Streaming Features (OSF) is a data streaming scenario, in which the number of instances is fixed while feature space grows with time. This paper presents a rough sets-based online feature selection algorithm for OSF. The proposed method, which is called OSFS-NRFS, consists of two major steps: (1) online noise resistantly relevance analysis that discards irrelevant features and (2) online noise resistanlty redundancy analysis, which eliminates redundant features. To show the efficiency and accuracy of the proposed algorithm, it is compared with two state-of-the-art rough sets-based OSFS algorithms on eight high-dimensional data sets. The experiments demonstrate that the proposed algorithm is faster and achieves better classification results than the existing methods.
Eskandari, S. (2021). Streamwise feature selection on big data using noise resistant rough functional dependency. Journal of Mathematical Modeling, 9(4), 677-690. doi: 10.22124/jmm.2021.18707.1603
MLA
Sadegh Eskandari. "Streamwise feature selection on big data using noise resistant rough functional dependency". Journal of Mathematical Modeling, 9, 4, 2021, 677-690. doi: 10.22124/jmm.2021.18707.1603
HARVARD
Eskandari, S. (2021). 'Streamwise feature selection on big data using noise resistant rough functional dependency', Journal of Mathematical Modeling, 9(4), pp. 677-690. doi: 10.22124/jmm.2021.18707.1603
VANCOUVER
Eskandari, S. Streamwise feature selection on big data using noise resistant rough functional dependency. Journal of Mathematical Modeling, 2021; 9(4): 677-690. doi: 10.22124/jmm.2021.18707.1603