EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
3051. Multi-Purchase Choice Models of a Sequential Purchasing Process
Invited abstract in session WD-61: Online, Omnichannel, and Pricing, stream Retail Operations.
Wednesday, 14:30-16:00Room: S10 (building: 101)
Authors (first author is the speaker)
1. | Kexin Lai
|
Mathematical Sciences, University of Souhampton | |
2. | Christine Currie
|
School of Mathematics, University of Southampton | |
3. | Selin Ahipasaoglu
|
University of Southampton |
Abstract
We consider the situation where customers make a finite sequence of purchase decisions and buy at most one item at each stage, where the sets of items on offer at each stage are disjoint and could be drawn from one or more categories. Our aim is to optimize item prices to maximize overall expected revenue. The basis of the optimization is a choice model that describes how customers make purchase decisions at each stage. We consider two choice models: a single-stage bundle-purchase model (known as the Multi-variate Multinomial Logit model) and a multi-stage sequential purchase model.
Under the single-stage model, we assume that customers form a choice ‘bundle’ by selecting at most one item from each product set. We model this behavior similarly to the standard multinomial logit model, where customers choose the utility-maximizing bundle. In the multi-stage sequential choice model, customers make a sequence of choices, where the choice made in one stage depends on the choices made in the previous stages. We include product interactions in both models to account for product compatibility.
Similar problems exist in industries such as tourism (travelers visit different destinations), food (restaurants with a rotating selection of specials), retail (stores with rotating inventory), and entertainment (cinemas with rotating films). Our empirical results on real data suggest that models with greater purchase history and information attain higher levels of accuracy.
Keywords
- Revenue Management and Pricing
- Marketing
Status: accepted
Back to the list of papers