Optimal product line design using Tabu Search

Product design constitutes a critical process for a firm, which if not implemented effectively it may even question its viability. The optimal product line design is an NP-hard problem, where a company aims at designing a set of products that will optimize a specific objective. Whilst Tabu Search (TS) has effectively solved a large number of combinatorial optimization problems, it has not yet been evaluated in product design. In this paper we design and implement a TS algorithm, which is applied to both artificial and actual consumer-related data preferences for specific products. The algorithm’s performance is evaluated against previous approaches like Genetic Algorithm and Simulated Annealing. The results indicate that the proposed approach outperforms nine tested heuristics in terms of accuracy and efficiency. It also constitutes a more robust technique, and can be effectively generalized to larger problem sizes, which include higher number of products, attributes, or levels. Finally, a novel variant of TS capable of reducing execution time called Tabu Search Class Move, is introduced.