Solving Problems in Industry - Java Data Mining: Strategy, Standard, and Practice - page 25

Java Data Mining (JDM) is geared toward providing a rich set of
capabilities that allow applications to integrate data mining pro-
cesses and results to solve a wide variety of problems. As such, JDM
fits well with a solution strategy involving data mining.

2.1 Cross-Industry Data Mining Solutions

Data mining can solve a wide variety of problems either to gain
understanding or insights from business and scientific data (what is
the root cause of failures), or to predict certain outcomes (will the
customer purchase the product?) or values (what is the predicted real
estate value?). Knowing how to adapt even a small set of canonical
problems to specific industries enables taking advantage of the
power of data mining. This section explores several cross-industry
solutions where data mining is being applied. We use the term
“cross-industry” since most industries can tailor each of these solu-
tions for their particular domain.

2.1.1 Customer Acquisition

Obtaining new customers is the hallmark of growth for many
companies. However, not all customers are equally profitable. Busi-
ness marketers may select a subset of customers using some basic
criteria, such as income and age constraints, but may prove to be not
loyal, that is, not stay with the company for a long time or purchase
products exclusively from that company. Providing generous offers
to attract such customers may result in high acquisition costs while
providing no long-term benefits. Some customers simply jump from
offer to offer to take advantage of discounts or “freebies.”
Others may be loyal, but may purchase infrequently or purchase
only low-margin products. Targeting low-value customers with
generous offers may also prove to be counterproductive. Targeting
the right potential customers can result in greater customer retention,
greater customer lifetime value, and more profitability in the most
desirable product and customer segments.
Consider the customer groups in Figure 2-1. A potential customer
population for a marketing campaign is represented by the large
oval. A company could attempt to acquire all these customers, but
that may be prohibitively expensive. Moreover, many of the potential
customers may quickly leave for a competitor, purchase only low-
margin products, or introduce more costs (through call center or