Data Mining Cookbook: Modeling Data for Marketing, Risk, and Customer Relationship Management - Softcover

Über die Autorin bzw. den Autor

OLIVIA PARR RUD (Olivia@datasquare.com) is Executive Vice President of Data Square, LLC, a leading database marketing consulting firm. She has over 22 years' experience in data mining, predictive modeling, and segmentation for a variety of industries, including credit card, insurance, high tech, telecommunications, and catalog industries. She provides analysis and solutions for her clients in the areas of acquisition, retention, risk, and overall profitability for direct mail, telemarketing, broadcast marketing, and the Internet.

Von der hinteren Coverseite

Get a data mining recipe for success to increase profits and reduce costs in today's business world with-This book shows you how to create and implement models of the most commonly asked data mining questions for marketing, sales, risk analysis, and customer relationship management and support. You'll get proven modeling techniques that address specific questions to help you find new ways to increase profit and cut costs. Starting from the basics, you'll learn how to plan the menu and choose the right ingredients-or ask the right questions and get data ready to mine-before you get down to the business of creating the meal. You'll find numerous case studies that detail available data sources for developing targeting models, then learn to process, evaluate, and implement them through an extensive case study of a lifetime value model for a life insurance direct-mail campaign. This step-by-step guide will help you to:
* Mine your company's data or find outside sources for your project
* Select and transform the variables when preparing data for modeling
* Test and validate response, activation, and profitability models
* Perform customer analysis through profiling and segmentation
* Build models to predict response, risk, churn, and lifetime value
* Use models for Web-based marketing and customer support
 
Also AvailableData Mining Cookbook CD-ROM see ad in back of bookVisit our Web site at www.wiley.com/compbooks/ Visit the companion Web site atdataminingcookbook.wiley.com

Auszug. © Genehmigter Nachdruck. Alle Rechte vorbehalten.

The Data Mining Cookbook : Modelling Data for Marketing, Risk and Customer Relationship Management

By Rud, Olivia Parr

Chapter 1: Setting the Objective

In the years following World War II, the United States experienced an economic boom. Mass marketing swept the nation. Consumers wanted every new gadget and machine. They weren't choosy about colors and features. New products generated new markets. And companies sprang up or expanded to meet the demand.

Eventually, competition began to erode profit margins. Companies began offering multiple products, hoping to compete by appealing to different consumer tastes. Consumers became discriminating, which created a challenge for marketers. They wanted to get the right product to the right consumer. This created a need for target marketing-that is, directing an offer to a "target" audience. The growth of target marketing was facilitated by two factors: the availability of information and increased computer power.

We're all familiar with the data explosion. Beginning with credit bureaus tracking our debt behavior and warranty cards gathering demographics, we have become a nation of information. Supermarkets track our purchases, and Web sites capture our shopping behavior whether we purchase or not! As a result, it is essential for businesses to use data just to stay competitive in today's markets.

Targeting models, which are the focus of this book, assist marketers in targeting their best customers and prospects. They make use of the increase in available data as well as improved computer power. In fact, logistic regression, which is used for numerous models in this book, was quite impractical for general use before the advent of computers. One logistic model calculated by hand took several months to process. When I began building logistic models in 1991, 1 had a PC with 600 megabytes of disk space. Using SAS, it took 27 hours to process one model! And while the model was processing, my computer was unavailable for other work. I therefore had to use my time very efficiently. I would spend Monday through Friday carefully preparing and fitting the predictive variables. Finally, I would begin the model processing on Friday afternoon and allow it to run over the weekend. I would check the status from home to make sure there weren't any problems. I didn't want any unpleasant surprises on Monday morning.

In this chapter, I begin with an overview of the model-building process. This overview details the steps for a successful targeting model project, from conception to implementation. I begin with the most important step in developing a targeting model: establishing the goal or objective. Several sample applications of descriptive and predictive targeting models help to define the business objective of the project and its alignment with the overall goals of the company. Once the objective is established, the next step is to determine the best methodology. This chapter defines several methods for developing targeting models along with their advantages and disadvantages. The chapter wraps up with a discussion of the adaptive company culture needed to ensure a successful target modelling effort.

Defining the Goal

The use of targeting models has become very common in the marketing industry. (In some cases, managers know they should be using them but aren't quite sure how!) Many applications like those for response or approval are quite straightforward. But as companies attempt to model more complex issues, such as attrition and lifetime value, clearly and specifically defining the goal is of critical importance. Failure to correctly define the goal can result in wasted dollars and lost opportunity.

The first and most important step in any targeting-model project is to establish a clear goal and develop a process to achieve that goal. (I have broken the process into seven major steps; Figure 1.1 displays the steps and their companion chapters.)

In defining the goal, you must first decide what you are trying to measure or predict. Targeting models generally fall into two categories, predictive and descriptive. Predictive models calculate some value that represents future activity...