An incremental approach for calculating dominance-based rough set dependency

Feature selection and classification are widely used in machine learning in the context of big data. In many data sets, both attributes and decision classes can be preference ordered. Therefore, to process the data and information based on preference-ordered attributes, dominance-based rough set approach (DRSA) has been proposed. DRSA considers dominance relation between objects and can process the information with preference-ordered attribute domains. The it should be noted that the majority of the algorithms based on DRSA use dependency as an underlying criterion measure for different tasks. However, calculating dependency using the conventional DRSA approach requires the calculation of lower and upper approximations which is a computationally expensive task. A new approach has been proposed in this paper which calculates the dominance-based rough set dependency measure without calculating the lower and upper approximations. The proposed methodology is called the “Incremental Dominance-based Dependency Calculation Method” (IDDC). To justify the proposed approach, both IDDC and conventional approaches are compared using various data sets from the UCI data set repository. Results have shown that the proposed approach outperforms the conventional approach by depicting on average 46% and 98% decrease in execution time and required runtime memory, respectively.