226 Part II • Information Technology
making the data available to everyone, whereas the analysis
of the data is performed by and/or for a single manager or a
small group of managers and is, therefore, a managerial
support system. Without explicitly mentioning it, we have
already begun the more detailed discussion of these tools for
analyzing data in the warehouse, for the DSSs described in
the previous section often pull the data they need directly
from the organizations’ data warehouses.
Data miningemploys a variety of technologies
(such as decision trees and neural networks) to search
for, or “mine,” “nuggets” of information from the vast
quantities of data stored in an organization’s data
warehouse. Data mining, which is sometimes considered a
subset of decision support systems, is especially useful
when the organization has large volumes of transaction
data in its warehouse. The concept of data miningis not
new, although the term became popular only in the late
1990s. For over two decades, many large organizations
have used internal or external analysts, often called man-
agement scientists, to try to identify trends, or patterns, in
massive amounts of data by using statistical, mathematical,
and artificial intelligence (AI) techniques. With the devel-
opment of large-scale data warehouses and the availability
of inexpensive processing power, a renewed interest in
what came to be called data mining arose in recent years.
Along with this renewed interest came a variety of
high-powered and relatively easy-to-use commercial
data mining software products. Among these products
are IBM SPSS Modeler Professional, Oracle Data
Mining, Salford Predictive Miner, SAS Enterprise Miner
and Text Miner, TIBCO Spotfire Miner, XLMiner for
Windows (an add-in for Microsoft Excel from Resampling
Stats), and KnowledgeSEEKER, KnowledgeSTUDIO, and
StrategyBUILDER from Angoss Software (based in
Canada). Among the more interesting data mining
products are text mining products, such as SAS Text
Miner, which have the ability to handle textual informa-
tion, pulling data out of letters, memos, medical records,
blogs, wikis, tweets, and documents of all kinds and
finding themes and patterns in these documents. The data
mining products vary widely in cost, ranging from under
$1,000 for some desktop products to over $100,000
for some enterprise products that run on large servers.
Consultants are often required to fully utilize the capabili-
ties of the more comprehensive products.
What are the decision techniques or approaches used
in data mining? One key technique, decision trees, is
embedded in many of the packages. A decision tree is a
tree-shaped structure that is derived from the data to
represent sets of decisions that result in various outcomes—
the tree’s various end points. When a new set of decisions is
presented, such as information on a particular shopper, the
decision tree then predicts the outcome. Neural networks, a
branch of artificial intelligence to be discussed later in this
chapter, are incorporated in most of the high-end products.
Other popular techniques include linear and logistic regres-
sion; association rules for finding patterns of co-occurring
events; clustering for market segmentation; rule induction,
the extraction of if-then rules based on statistical signifi-
cance; nearest neighbor, the classification of a record based
on those most similar to it in the database; and genetic
algorithms, optimization techniques based on the concepts of
genetic combination, mutation, and natural selection.
For completeness, let us introduce a term related to
data mining, but with a difference—online analytical
processing (OLAP). OLAP has been described as human-
driven analysis, whereas data mining might be viewed as
technique-driven. OLAP is essentially querying against a
database, employing OLAP software that makes it easy to
pose complex queries along multiple dimensions, such as
time, organizational unit, and geography. The chief
component of OLAP is the OLAP server, which sits
between a client machine and a database server. The
OLAP server understands how data are organized in the
database and has special functions for analyzing the data.
In contrast, data mining incorporates such techniques as
decision trees, neural networks, and genetic algorithms.
An OLAP program extracts data from the database and
structures it by individual dimensions, such as region or
dealer. Data mining software searches the database for
patterns and relationships, employing techniques such as
neural networks.
Of course, what you can do with data mining is more
important to you as a manager than the decision techniques
employed. Typical applications of data mining are outlined
in Table 6.1. Whatever the nature of your business, thesimulation, and exact and heuristic algorithms, it designs efficient floor plans for the POD facility and
determines the staffing needed and their placement within the facility. RealOpt-RSS is a DSS for the efficient
management of the logistics of receipt, stage, and storage (RSS) facilities and regional distribution nodes
for medical countermeasures. Finally, RealOpt-CRC is concerned with radiological emergency planning
and response. Since 2005, RealOpt has been successfully used in planning exercises to respond to
simulated pandemic events and bioterrorist attacks at various locations in the United States.A Potpourri of DSS Examples (continued)