Network revenue management under speci c choice models

New challenges in the business environment, such as increased competition and the in uence of the Internet on main distribution channels has led to fundamental changes in traditional revenue management models. Under these conditions, modeling individual decisions more accurately is becoming a key factor. Nearly all research studies about choicebased revenue management models use the well-known multinomial logit model. This model has one important restriction, that is, the independence of irrelevant alternatives; a property which states that the ratio of choice probabilities for two distinct alternatives is independent of the attributes of any other alternatives. In this paper, a nested logit model is proposed for removing this limitation and incorporating a correlation between alternatives in each nest. The new subproblem of column generation is introduced and a combination of heuristic and metaheuristic algorithms for solving this problem is provided. Interesting outcomes are obtained during analysis of the results of experimental computations, such as oer sets and iteration trends, with respect to the correlation measure inside each nest. Simulation results show that, although changing the choice model might lead to signicant improvement in revenue under some conditions, during all scenarios, observing the correlation should not cause the choice model to change immediately.