Mining Your Own Business in Retail Using DB2 Intelligent Miner for Data

Chapter 4. How can I characterize my customers from the mix of products that they purchase? 93

In the following sections we look at a couple of examples of how customer

scoring can be used.

4.7.2 Using the cluster results to score all your customers

The usual application for deploying the cluster/segmentation results is where,

following an analysis of the data derived segments, you want to be able to

classify your customers in particular segments and use this information in some

specific way. A number of campaign management tools (for example, Xchange

Campaign) require you to provide a table of customers and the segments to

which they have been assigned. Alternatively, you may want to use the segments

as additional dimensions in some type of On-Line Analytic Processing (OLAP)

tool and perform some more detailed investigation of the customer attributes. In

this case a similar table is required.

The simplest way to generate this type of table is to use the first of the two

methods described above. This method assumes that the data mining tool has

direct access to the appropriate customer records and that these records are

stored either as a single table, or an equivalent database “view”, at which the

data mining tool can be directed. Using the application mode of the data mining

tool then enables all customers to be scored and assigned to a segment with the

results being written back the another database table or file. This is illustrated in

Figure 4-19.