A probabilistic approach to setting weights in a weighted-average product evaluation

Abstract

This paper presents a procedure that estimates how markets perceive a product. A weighted-average approach is used as a model for the product perception. Since this approach requires weights to be assigned to product features, and the weights are unknown and vary among customers, the procedure focuses on estimating expected weights of the entire market such weights are representative weights. The estimation is performed by constructing a confidence interval for the weights. Further, a more accurate location of the weights in the interval is proposed, the accuracy being based on additional information regarding inferior products. This information suggests that if a product is inferior, its value must be low. Therefore, the weights from the interval which minimize the value of an inferior product are suggested, the minimization serving as an approximation of what low value means. Since a minimization is involved, its existence and uniqueness is discussed.