Product Line Design Search Considering Reliability and Robustness Under Uncertainty in Discrete Choice Methods

Point-estimates of part-worth values in customer preference models have been used in market-based product design under the simplying assumption that customer preferences can be treated as deterministic. However, customer preferences are not only inherently stochastic, but are also statistical estimates that exhibit random errors in model formulation and estimation. Ignoring uncertainty in customer preferences and variability in estimates has caused concern about the reliability and robustness of an optimal product design solution. This study quantitatively defines reliability and robustness of a product design under uncertainty when using discrete choice methods. These metrics are then integrated into a multi-objective optimization problem to search for product line solutions considering reliability and robustness under uncertainty when using discrete choice methods.