DBMS - Data Mining Solutions 1998

The DBMS Guide to
Data Mining Solutions

Data Mining is moving beyond the early adopter stage. Are you ready?

As data mining enters the mainstream, an in-depth understanding of data mining techniques and applications will spell the difference between success and failure. Drilling down, pivoting, and other OLAP techniques are intuitive and easy to grasp. But analysts applying data mining must be well versed in the variety of techniques available, the kinds of analyses for which each technique is appropriate, how data must be prepared, how each technique operates, and how to interpret the results properly. Thatıs why DBMS produced this special supplement about data mining applications, methods and tools. We hope you find it useful! Your feedback is always welcome. Please write to dbms@mfi.com.

ı Maurice Frank, Editor-in-Chief

Contents

Data Mining and Knowledge Discovery
Data mining is poised to move beyond the early adopter stage to a much wider audience. But what kinds of business problems can data mining technology solve, and what must users understand to apply these tools effectively? This introduction surveys popular data mining techniques, the knowledge discovery process, and how data mining can solve business analysis problems. The authors also examine future trends affecting the data mining market.

Predicting Credit Risk
One challenge data miners face is choosing which technique to apply to an analysis task. In many situations, analysts can use a variety of techniques, but each technique conducts the analysis and represents the results differently. This sidebar describes sample data for a credit risk classification task that each data mining technique can address.

Association and Sequencing
Selling as many different products as possible to your customers maximizes their value to your business. Association and sequencing techniques, also known as market basket analysis, can reveal purchasing patterns such as which products are purchased at the same time or which products are purchased after other products. Pattern detection has many other applications in finance and other business processes.

Classification and Regression
Effective data mining can deliver a high payoff for your business. Classification and regression can predict customer behavior, signal potentially fraudulent transactions, and forecast store profitability, to name just a few of many applications. This article introduces four classification and regression techniques: decision trees, neural networks, Naıve-Bayes, and nearest neighbor.

Decision Trees
When a businessperson needs to make a decision based on several factors, a decision tree can help identify which factors to consider and how each factor has historically been associated with different outcomes of the decision. Data mining tools that produce decision trees create graphical models and text rules that can both predict behavior and explain patterns. End users can easily understand and apply decision trees, so these techniques are very popular data exploration tools.

Neural Networks
Have you ever made an extraordinary purchase on a credit card only to be embarrassed when the charge wasnıt authorized? Somehow your transaction was flagged as possibly being fraudulent, and a neural network could have been involved. Neural networks create models that can rapidly analyze large amounts of data and make predictions in real time. But because they can't explain how a prediction was made, many end users consider these sophisticated techniques a mysterious black box.

Naıve-Bayes and Nearest Neighbors
This pair of easy-to-use techniques can quickly build classification models for prediction and description of data. However, data miners must understand their limitations before applying these techniques.

Buyer's Guide
More than 20 data mining products.

Data Mining Resources on the Web

About the Authors
Estelle Brand (estelle@xore.com) and Rob Gerritsen (rob@xore.com) are founders of Exclusive Ore Inc., based in Bluebell, Pennsylvania, which is a consulting and training company specializing in data mining. During the last two years they have used more than a dozen data mining products. Their database management systems experience dates back to the dark ages. For more information about Exclusive Ore and data mining, see www.xore.com.

Please send questions or comments to dbms@mfi.com
Updated July 8. 1998.