Data Intelligence, Business Analytics
A Classification and Regression Tree (CART) datamining technique can be used to analyze a website's home page clickthrough data to drive its redesign to appeal to target audiences. Let's say that a winery, Vincor International, has purchased a banner ad on the http://www.vancouversun.com/ home page. The campaign's objective is to drive as many people as possible to the http://www.vincorinternational.com/ website to buy wine from, or register for a wine tour at, one of Vincor's many Canadian winery locations.
Every website campaign has an intended market audience, and in the case of a winery, professional females between the ages of 34 and 65 are an important target. The Classification and Regression Tree (CART) algorithm in the http://www.angoss.com/ datamining tool can help to determine what elements on the Vincor website appeal the most to its target market. Findings from this exercise might drive design improvements to the website that appeal more to Vincor's female, middle-aged professional audience.
Conducting this analysis requires that Vincor's website have a back-end web analytics package to capture information about website visitors: a free http://www.google.com/analytics/ package or a high-end web analytics package from http://www.webtrends.com/. Regardless, the web analytics package must capture demographic data about each visitor such as age, location of residence, gender, household income, clickthrough data for the Vancouver Sun banner ad, and Vincor website home page clickthrough data.
For this analysis, the target, dependent variable can be 'number of clickthroughs' or '% of clickthroughs' in total from the Vincor home page to other areas of the site. Independent variables might include Vincor website visitor demographic data, a mutlivariate variable with categories representing home page clickthrough elements, and a binary variable indicating whether a visitor arrived via the Vancouver Sun home page banner ad.
Analysis of this data using a Classification and Regression Tree (CART) datamining technique can help to determine what content and areas of Vincor's website must be changed in order to maximize online wine purchases and wine tour registrations on its site.