SQL Server Data Mining—Approaching the Customer-Segmentation Problem

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

Approaching the Customer-Segmentation Problem

Because users can slice and dice information by all the attributes in a dimension rather than just predefined drilldown hierarchies, analysts could use the new Internet-related attributes that we added to drill down through the data and start to understand how customers' online activities affect measures such as total sales or profitability. For example, they can learn that frequent visitors to the site often have high sales amounts, but that this isn't always the case—some frequent visitors are "just looking."

To really do a good job of targeting the DVD marketing campaign to customers likely to act on the information, analysts need to perform a segmentation exercise where all customers that have similar attributes are categorized into groups. Because the list of customers is huge and there is a large number of attributes, we can start this categorization process by using a data mining algorithm to search through the customers and group them into clusters.

The Microsoft Clustering algorithm is a great tool for segmentation and works by looking for relationships in the data and generating a list of clusters, as shown in Figure 10-4, and then gradually moving clusters around until they are a good representation of the data.

Figure 10-4 Clusters of data