A rough set-based Competitive Intelligence approach for anticipating competitor’s action

Abstract

In an evolving competitive environment, characterized by an increasing competition and a rapid market change, companies strive to reach their competitive advantage by monitoring and processing the information related to such environment. Getting better informational and intelligence support is consequently critical and vital. Competitive Intelligence (CI) appears then as a vital component of strategic planning and management process within organization. It has attracted increasing attention over the past two decades and has become a challenge for business professionals and begun to be studied and implemented in large and small companies, in the private and public sectors, and within different industrial contexts. Interest towards CI is observed as well among academic researchers. Existing literature reveals that CI is a multi-disciplinary concept studied by researchers with different fields of expertise (management, marketing, information technology, decision support, and knowledge management) and addressed from different points of view (as concept, product, process, practice/discipline, method, and as a system). Despite this great diversity in the body of knowledge related to the CI, there is no empirical study that touched the practical aspect of the problem by developing a complete CI solution that can be delivered to the decision maker in terms of anticipating competitors’ reactions. Existing research works focus more on the process, its steps, objectives and benefits for the company. Our paper aims to address this gap and present the first practical end-to-end CI solution based on rough set theory to anticipate competitor’s decisions. To achieve this objective, several theoretical contributions were reached. Our proposed approach consists of four phases: (i) actions collection; (ii) a two-step actions association in which first we cluster actions using k-modes algorithm then we proceed to associate the competitive actions using our new proposed algorithm for action association,; (iii) rules generation using rough set theory with a modified version of LEM2 algorithm which can deal with inexact data and derive rules from it; and finally (iv) rules aggregation using a new proposed algorithm. As motivation for the research, we use a real case study in telecommunications to illustrate our whole proposed approach and show its efficiency.