Feature extraction using rough set theory in service sector application from incremental perspective

Abstract

In service industry application, there is vague and qualitative information required to be processed properly, for example, to identify customer preferences in order to provide adequate services. From literature, Rough Set Theory (RST) has been indicated to be one of promising approaches to cope with vagueness in a large scale database. Basically, the rough set approach integrates learning-from-example techniques, extracts rules from a data set of interest, and discovers data regularities. Most of the existing RS based approaches are able to implement rule induction but it is very time consuming from computation perspective particularly from a large database. To date, there is a demand to generate and analyze business decision rules based on dynamical data sets and conclude such rules on the daily basis in the service industry. Therefore, in this study, an Incremental Weight Incorporated Rule Identification (IWIRI) algorithm is proposed to fulfill such demand. The proposed approach is proficient to efficiently process in-coming data (objetcs) and generate updated decision rules without re-computation efforts in the database. Identification of features based on the customer's preference and implementation of the proposed algorithm are summarized in the case study. This paper forms the basis for solving many other similar problems that occur in service industries.

Binary Representation of Polar Bear Algorithm for Feature Selection

“…Li et al [15] proposed a new method based on the Wolf Search Algorithm (WSA) for optimizing the feature selection problem. Huang et al [16] introduced an Incremental Weight Incorporated Rule Identification (IWIRI) algorithm to process in-coming data (objects) and generate updated decision rules without re-computation efforts in the database. Alia et al [17] developed a new algorithm for Feature Selection based on hybrid Binary Cuckoo Search and rough set theory for classification on nominal datasets.…”