Airlines have routinely used current and historical demand to update their prices and their allocation of seat inventory to different fare classes. This process involves extensive computation, including demand forecasting, modeling of csustomer behavior, and optimization. Similar problems arise in the pricing of other resource-intensive, time-definite services, such as allocation of bandwidth in telecommunications. Other industries have begun to explore analogous methods for pricing and allocating products and services. (See [Dana (1998), Dana (1999), Mahajan and van Ryzin (2001), Mahajan and van Ryzin (2001)].) Software firms provide decision support tools that aid in determining retail markdowns (when and by how much to mark down merchandise) and category pricing (if you manage a range of close substitutes how do you set prices on each). Integrated into the underlying mathematics are models of buyer behavior. Some of these models assume myopic buyers who make a buy/no-buy decision based on price, and do not consider the option of waiting for a price cut. There is a sizable literature in economic theory that considers pricing in an environment where buyers are forward-looking; however, these models have not been applied to support commercial decision making and this workshop will consider them as well as the corresponding challenges for computer science. Major challenges here lie not in code but in the need for new algorithmic ideas that will allow the use of more sophisticated models of buyer behavior.
Until recently yield-management by airlines was done by individual flights or legs, with limited representation of substitutability. Recently, origin-destination methods, which consider multiple routes between the same pair of cities, have been made available by PROS. Capturing these substitution effects required significant, domain-specific modeling, additional input data about schedules and customer preference, and increased computation. Analogous substitution and bundling occurs in other industries such as retail, telecomunication, other travel and transport, but our understanding of customer preference is very limited. (See McGill and van Ryzin (1999), Talluri and van Ryzin (1998), Talluri and van Ryzin (1999)].) In telecomunications or package transport, this may not matter as the customer cares only that the call is completed or the item is transported, not about the specific route that it follows. However, in retail, travel, healthcare, and other services where the composition of the offering is visible to the end customer, a far better understanding of customer preferences is needed. Modeling of bundling and substitution requires an understanding of the physical items and processes under consideration. Incorporating preference for specific items and/or combinations requires an understanding of customer behavior. Methods for systematically using vast amounts of transaction data from sources such as web-sites, point-of sale-terminal, customer surveys, and the like, to build models of customer behavior, remains a significant computational challenge, and one this workshop will investigate. A related issue is in robustness of assumptions about the distribution of demand. This is a natural environment for on-line analysis, again an area of considerable interest to computer scientists.
Dana, J., "Using yield management to shift demand when the peak time is unknown'', The RAND Journal of Economics, Fall 1999, 456-474.
Mahajan, S. and van Ryzin, G.J., "Inventory competition under dynamic consumer substitution,'' Operations Research, 49, 2001, 646-657.
Mahajan, S. and van Ryzin, G.J., "Stocking retail assortments under dynamic consumer substitution,'' Operations Research, 49, 2001, 334-351.
McGill, J. and van Ryzin, G.J., "Revenue management: Research overview and prospects,'' Transportation Science, 33, 1999, 233-256.
Talluri, K. and van Ryzin, G.J., "An analysis of bid-price controls for network revenue management,'' Management Science, 44, 1998, 1577-1593.
Talluri, K. and van Ryzin, G.J., "A randomized linear programming method for computing network bid prices,'' Transportation Science, 33, 1999, 207-216.