|•Data mining enables managers to sift through
data in a number of ways. Each method produces different information that
managers can then base their decisions on.
|•The following are five things managers do to
make their data meaningful:
|1.Classification. Before mining, managers
define data classes that they think will be helpful in spotting trends. They
then apply these class definitions to all unclassified data to prepare it for
|2.Estimation. When managers classify data,
the record either fits the classification criteria or it doesn’t. Estimation
enables managers to assign a value, based on some criterion, to data. For
example, assume a bank wants to send out credit card offers to people who are
likely to be granted a credit card. The bank may run the customers’ data
through a program that assigns them a score based on where they live, their
household income, and their average bank balance. This provides managers with
an estimate of the most likely credit card prospects so that they can include
them in the mailing.
|3.Affinity grouping or association rules.
When mining data, managers can also determine which data goes together. In
other words, they can apply affinity grouping or association rules to the
data. For example, suppose analysis of a sales database indicates that two
items are bought together 70 percent of the time. Based on this data,
managers might decide that these items should be pictured on the same page in
the next mail-order catalog they send out.
|4.Clustering. Clustering involves
organizing data into similar subgroups, or clusters. It is different from
classification in that there are no predefined classes. The data-mining
software makes the decision about what to group together, and it is up to
managers to determine whether the clusters are meaningful. For example, the
data-mining software may identify clusters of customers with similar buying
patterns. Further analysis of the clusters may reveal that certain
socioeconomic groups have similar buying patterns.
|5.Description and visualization. Often,
the purpose of data mining is merely to describe data so managers can
visualize it. Sometimes having a clear picture of what is going on with the
data helps people to interpret it in new and different ways.