The 2011 ENBIS Challenge
(Sponsored by JMP)
Maximising Click Through Rates on Banner Adverts:
Predictive Modeling in the On Line World
The Internet has afforded new ways to monetise products and services through its capability to address individual consumers and businesses directly and immediately. This year's ENBIS Challenge is about the application of predictive modeling, also known colloquially as ‘data mining’, to a specific on-line setting.
An online portal (‘Content4U’) gives access to news and lifestyle content. Although users are not charged for accessing this content, there is no ‘free lunch’, so money has to be made somehow. To do this, the portal places advertising within the free content, and charges the advertisers for the privilege: When a user selects a particular advert, they are transferred to the merchant’s site to learn more about the offer, and at the same time, the advertiser pays the portal for providing a lead. By increasing the click through rate to a merchant’s site the portal will make more money and will be able to finance or even expand its own services.
The challenge consists of attempting to increase advertising revenues to the portal by dynamically surfacing specific advertisements that particular types of people will find attractive and are more likely to click on. There is a body of observational data that describes both the profile of a user and what free content they have viewed in the past, so the hope is that we can build a useful model that will predict the profile of a user from their viewing habits alone. If we succeed, we can then display appropriate,personalised adverts. Note that for good reasons, people are generally reluctant to give personal information to an on-line portal, so the observational data was specially collected with this modeling in mind. Read more ....