226 Enhance Your Business Applications: Simple Integration of Advanced Data Mining Functions
Example 10-3 Extracting error rate from test result
select ID, IDMMX.DM_getClassError(result) as ErrorRate
10.5 Hybrid modeling
The DB2 data mining function of IM Modeling offers ease of use for several
mining techniques. It also offers ease of use when you want to apply hybrid
modeling (see also 1.2, “Data mining does not stand alone anymore” on page 6).
Hybrid modeling is a clear feature of IM Modeling. Hybrid modeling is practiced
when you want to apply several mining techniques in sequence to the data.
For example, in case of a predictive modeling of a debit rating of banking card
holders, Figure 10-1 uses demographical clustering to profile each cluster. This
way you can characterize each customer according to the cluster to which it
belongs. From this, you get several homogenous groups of customers that are in
our population of banking card holders.
Next you select a number of the clusters that have profiles which interest you
most according to the business issue. In the second stage of modeling, the
decision tree and its rules tell you in which nodes the customers of a certain
cluster belong. These rules help to generate more lift when you target the
customers of each cluster for a competitive debit rating campaign.