Incremental attribute reduction with rough set for dynamic datasets with simultaneously increasing samples and attributes

Attribute reduction with rough set is a popular data analysis methodology for data dimensionality reduction. For dynamic datasets, the existing research has mainly focused on incremental attribute reduction with increasing samples (rows) or attributes (columns), but there is hardly any further research on attribute reduction for dynamic datasets with simultaneously increasing samples and attributes. This paper presents a novel incremental algorithm for attribute reduction with rough set. Firstly, the definition of discernibility relation is proposed based on the improved discernibility matrix. Then, the incremental mechanisms of samples and attributes are studied in terms of discernibility relation under a unified framework. On the basis of two incremental mechanisms, a unified incremental mechanism is introduced for dynamic datasets with simultaneously increasing samples and attributes, and the incremental algorithm is developed according to the unified incremental mechanism. The proposed algorithm has the solid mathematical foundation, which is also suitable for datasets with massive samples and attributes. Finally, compared experimentally with other algorithms, the efficiency of the developed incremental algorithm is demonstrated in terms of running time.