A FRAMEWORK FOR ATTRIBUTE SELECTION IN MARKETING USING ROUGH COMPUTING AND FORMAL CONCEPT ANALYSIS

Convergence of computers and information technology has changed the real life of the common man in many ways due to accumulation of data and information. However, analysing the structured and unstructured dataset to arrive at decisions and identifying the chief characteristics affecting the decisions is a tedious task. Furthermore, knowledge extraction in uncertain data analysis and reduction of superfluous attributes is a major concern, which has not been taken into account in   prior research on decision making. Many researchers identified that rough set is a better “intelligent” technique for decision making. But, this technique is not a fail-safe in analysing quantitative information systems. To overcome this limitation, rough set on intuitionistic fuzzy approximation is hybridised with formal concept analysis, and a better model for decision making and identifying chief attributes affecting decisions is presented. The major advantage is that the model works for both qualitative and quantitative information systems. The proposed model uses two processes, pre-process and post-process. The pre-process identifies the “almost indiscernibility” among the attribute values and reduces the quantitative information system to qualitative information system by imposing ordering relations. Further, the reduced information system is processed and rules are generated using rough set. These decision rules are passed to formal concept analysis in post-process. The main objective of this process is to find the chief attributes affecting the decisions in the pre-process. If these attributes are properly addressed, better decisions could be achieved. To this end, this paper presents an empirical study on marketing to show the viability of the proposed research.