Data Mining with Microsoft® SQL Server® 2008

Chapter 18. Implementing a Web Cross-Selling Application

Cross-selling is a very common business problem. It involves suggesting a list of new products based on those in the customer's current or previous shopping basket. For example, if you go to Amazon.com and put a book in your shopping cart, you get a set of other book recommendations. This list is based on the market basket analysis of thousands of customers with similar purchases. Good recommendations improve customers' shopping experiences and, thus, increase the overall sales. Bad recommendations annoy customers and eventually drive them away. The major challenge of cross-selling is how to give each customer the right set of recommendations. If the shop product catalog is small, it is relatively easy to give suggestions based on marketing experiences. However, when the number of distinct products is large, the problem is pushed to a new dimension.

Suppose that you are the owner of MovieClick.com, a fictional online retail store that sells movies. You have thousands of movies at MovieClick.com. You want to increase movie sales by giving online shoppers personalized suggestions. This chapter will help you solve this business problem using data mining techniques.

In this chapter, you will learn about the following:

• Source data descriptions

• Building recommendation models with the Microsoft Decision Trees and Microsoft Association Rules algorithms

• The difference between Microsoft Decision Trees and Microsoft Association Rules for ...