SQL Server Data Mining—Asking the Right Question

Microsoft .NET Framework, ASP.NET, Visual C# (CSharp, C Sharp, C-Sharp) Developer Training, Visual Studio

Asking the Right Question

The best way to successfully set up a sensible data mining model is to be precise about the question you ask. Because we are looking for DVDs that sell well together, is the question we are asking "Which other DVDs have been bought during the same shopping trip?" or rather "Which other DVDs did customers also buy at some point?." If you were doing product recommendations on groceries, the first question would probably be the most sensible. The reason is that if someone is buying beer and wine today, we can probably recommend ice and potato chips because those are often sold in the same transaction.

However, in our case, we are trying to determine the likes and dislikes of consumers, which have a longer duration than just a single transaction. We are really trying to understand what kind of movies customers enjoy, so the second question is more appropriate for this business solution. To set up the model, we need to look at each customer and determine the list of DVDs that they have purchased. The data we are looking for looks something like Table 10-1. In data mining terminology, the customer would be the case, and the list of products for each customer would be a nested table.

Table 10-1. Customer DVD Purchase History