We have a project in the works to implement cross-sell on our web sites (and telephone orders of course) that is more focused on the merchandising group choose items that complement other items. However, during our data warehousing phase a couple years ago, we looked a number of god-awful expensive products that would use data-mining to extract product combinations that sell.
If you think about it, it isn't really rocket science. You process historical order data to produce a file of paired items that have been purchased together, then use it to generate statistics on which products are most likely to purchased together. Establish a cut-off point (number of sales/percentage of sales) to weed out the really weird combinations, then use the resulting file to recommend other products. (probably based on what you have on-hand) You could even create a process to build an index in real-time. You know, Diapers and beer. Anyway, I guess you could say I am interested. Warner Onstine wrote: > I have been banging around I an idea outlining how one would > duplicate/emulate what amazon does in relation to their recommendation > engine. > > Are there people on the list who want me to waste their time with this =)? > > -warner > > +warner onstine+ > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
