|M.Sc Student||Fligler Ariel|
|Subject||Product Line Design Using a Genetic Algorithm|
|Department||Department of Industrial Engineering and Management||Supervisors||Professor Ilan Shimshoni|
|Ms. Gila Fruchter|
In a marketing orientation era the problem of a product line design is a fundamental prerequisite for every business success. As customers are heterogeneous in preferences, the marketing concept calls for offering a product line instead of a single product. Given customers’ preferences and the cost of each possible product configuration, the manufacturer’s optimal product line design problem consists of finding the product and pricing strategies that maximize profits. The optimization problem becomes very complicated once the actual data for customer preferences is being used, and the classical optimization methods fail in finding a solution. We show that the usage of Genetic Algorithms, a mathematical heuristic mimicking the process of biological evolution, can efficiently solve the problem. Our genetic algorithm is extended with specially tailored problem domain operators and a linear programming post-processing step. Our method is applied to a laptop computer product line design with conjoint analysis data for customers’ valuations. Running the algorithm on simulated data that allows an analytical approach shows that our algorithm performs well on both data sets. Usage of domain specific operators and the usage of linear programming for local search have been found to be highly beneficial. Correlation has been found between products’ costs or customers’ reservation prices and the structure of the final solution. We show that products and customers that are correlated with higher utility values contribute higher profits to the manufacturer. We compare our algorithm’s results with a state of the art heuristic and show our algorithm’s dominance on both real and simulated data sets.