Marketing Decision Support Systems

University of Washington

Winter 1997

BBUS 429 B

Professor P.V. (Sundar) Balakrishnan

Office: Room 210

Tel. #: 685-5384

Email: sundar@u.washington.edu

Course Objectives:

Business is in the midst of a technological revolution. To be successful, students have to be at the forefront of this new technology. This means taking a "hands-on" approach to developing the requisite skills necessary.

This course deals with concepts, methods, applications of decision modeling to address various marketing issues. Unlike conventional capstone business courses that focus on conceptual material this course will attempt to provide skills to translate conceptual understanding into developing specific operational plans a skill in increasing demand in corporations today.

Using PC and UNIX-based computer software, students will develop Spreadsheet Models and Artificial Intelligence based Decision Support Systems for varying managerial decision contexts.

Specifically, course objectives are to:

• Provide you with understanding of the role of Artifical Intelligence, Expert Systems and Decision Models in managerial decision-making.
• Develop abilities to apply, build and modify decision models to solve real problems
• Explore the issues involved in the design and development of Artificial Intelligence Based Decision Support Systems and discuss the role these systems play in the business environment.
• Gain an In-Depth Knowledge of a particular type of Artificial Intelligence Technique, namely Genetic Algorithms.
• Gain the knowledge to build a prototype Artifical Intelligence Based Decision Support System.

The course will be of particular relevance to students planning careers in marketing and management consulting. The course is designed for students with some quantitative background as well as some exposure to marketing concepts.

REQUIRED COURSE MATERIAL

1) Gary L. Lilien, Marketing Management: Analytic Exercises for Spreadsheets, Second edition, 1993. (GL).

2) Various Readings; EXSYS Manual and Videotape placed on reserve in the library. (R).

RECOMMENDED REFERENCE

1) Gary L. Lilien, Philip Kotler, and K. Sridhar Moorthy, Marketing Models, Prentice Hall, 1992. (LKS).

This is a reference book that supplements course materials and class discussions. However, this is not a required text.

EVALUATION

Genetic Algorithms Term Project 35%

Assignments & Exams 35%

Class Presentations 15%

Class Discussions 10%

WORK LOAD

Class sessions will be devoted to probing, extending and applying the material in the readings and the cases. It is your responsibility to be prepared for each session as detailed in the course outline. Each one of you will benefit from belonging to a "study group" that meets and prepares for each session before coming to class.

The course will involve extensive computer-based work in addition to the readings and library research. My expectation is that the time required for out-of-class work will be 3 times the duration of class meeting.

Genetic Algorithms Term Projects Guidelines

1. Each group is expected to learn how to effectively search the Internet for information related to GA. In particular, each group will be expected to to download one of the "freeware" GA programs from various "sites" around the world and get it up and running on our computer systems (mainframes or PC). You will then provide a brief report (2 pages) evaluating this software.

1. Each Group is expected to develop a prototype Genetic Algorithms based Decision Support System (GADSS). The system must attempt to address a specific Marketing Problem.

You have a choice of working on your own problem and data set or working on developing a GADSS to address the problem and data set provided by the Instructor.

The GADSS can developed using the software provided (GENERATOR) or with the program that you have downloaded from another site.

A paper describing the developed systems as well as the diskette containing the developed system should be submitted.

Assignments & Exams:

There are various spreadsheet assignments (mini-cases) to be completed as part of the course. The dates on which the assignments have to be handed in are indicated in the course schedule. The exact nature of the assignments will be announced in class, sufficiently ahead of time. These are group assignments: Please form groups of two people to work on these assignments, and to prepare for class discussions.

1. All Software assignments that are to be turned in, must be wordprocessed and professional looking. NO handwritten work is permitted on the EXCEL spreadsheet assignments.
2. These Spreadsheet assignments are from your Text Book and can be done ahead of time.
3. You are expected to turn them in by end of Class-Period.
4. One or more of these assignments will be be earmarked as Exams and must be completed individually.
5. Groups will be expected to have worked on these AHEAD of time. You will be given some class time to complete your work. At this point, we will discuss these spreadsheet cases. A Group will be asked to come to the head of the class and demonstrate their working and proficiency of the underlying model. They will then moderate a discussion of that assignment. Every Group will get an opportunity to lead the discussion at least once.

Class Presentations:

This is one of the KEY's to doing well in this course. All readings as indicated in the Schedule must be done ahead of time. Each group (or sometimes individual) will be asked to present one of the readings each day. The presentations must be thorough and provoke discussions.

The quality of the discussions from the floor will help determine your grade for that day.

Class Participation:

Each of you is expected to contribute to class discussions. Do not expect to do well in this course by simply coming to class, taking notes, and synthesizing, recalling, or reproducing these notes for our evaluation. To do well, you must learn from active participation in class discussions.

In evaluating class participation, I will try to assess how your individual contribution enhance both the content and process of a discussion:

• do the comments add to our understanding of the issues or is it merely an attempt to get "air time" that day? (i.e. You have only has one thing to say, and want to say it no matter how irrelevant it is to the on-going discussion).
• are the comments timely and linked to the comments of others?
• do the comments move the discussion along by giving a new perspective?
• do your comments merely restate the facts in the reading materials and cases or do they provide new insights?
• do the comments reflect a concern for maintaining a constructive and comfortable classroom atmosphere?

If you are unprepared to participate in the day's discussions, notify me prior to the beginning of the class to avoid any embarrassment.

Software

We will play with a number of different software packages.

1. We will spend quite a bit of our time working with packages that deal with GENETIC ALGORITHMS. In your Computer Lab I have purchased and and had installed a package called GENERATOR, which is an EXCEL Spreadsheet add-on.
2. In addition, I will demonstrate GENESYS/GENELIN, a package that I have developed for Product Design.
3. Play with an Expert Systems Shell called EXSYS Pro (It is also limited to 50 rules. This is available in the machines in the Computer Lab) and a couple of Expert Systems for Marketing Decisions.

(You can also Purchase EXSYS Professional student version for $70. The Purchase can be done directly from the company EXSYS Inc. The phone number is 1-800-676-8356 or 505-256-8356. This is an expert system development package which is limited to 50 rules but has most of the functionality of the substantially more expensive EXSYS Professional Complete package. Additionally you will get the manuals, tutorials etc.)

1. In addition, the Text comes shrinkwrapped with a disk containing numerous cases requiring the use of EXCEL spreadsheet.

Paper Requirements

The paper should at a minimum discuss the following :

• The GADSS developed, features and shortcomings.
• The tool used for development and its evaluation.
• Further enhancements possible with the prototype developed.
• Discussions relevant to the project development scenario.
• Evaluation of the GA software that you have downloaded.

The following factors will be taken into consideration when grading the paper and the system:

• Creativity and features of the package utilized;
• Ease of interaction with the developed system;
• Validity of the developed system in terms of the queries asked by the system and the solution proposed;
• Complexity of the problem considered;
• Clarity of the paper;
• Structure and organization of the paper;
• Spelling and grammatical errors;
• Style of presentation - lucid or obscure;

Tentative Class Schedule (Revised)

Topic: Introduction

1/6 Search the Internet for GA Software

Topic: Basics of Spreadsheet Models

1/8 Read: Chapters 1, 9 {2 if you use LOTUS else skim} (GL)

Software: EXAMPLE

Topic: Tools & Shortcomings of Spreadsheet Models

1/13 Read: Chapters 1, 9 {2 if you use LOTUS else skim} (GL)

Software: EXAMPLE* (Due); HW Problem

Topic: Artifical Intelligence Techniques Overview

1/15 Read: "Moody's Evolving Desk" (R)

Software: GENERATOR

Topics: Decision Calculus Models: Fundamentals

1/22 Read: "Models & Manager: The Concept of Decision Calculus (R)

Read: "Commentary on Judgement based Marketing Decision Models", (R)

Read: "Decision Support Systems for Marketing Managers"(R)

Read: "Shake, Ratttle, & Roll" (R)

Software: ADBUDG (play); Estimating Parameters

Topic: Expert Systems Fundamentals

1/27 Read: Chapter 14 "AI, Expert Systems, and DSS" (R);

Read: Chapter B of EXSYS Manual (R)

Video: EXSYS Watch before class (Library Reference Desk)

Software: EXSYS DEMO (Lab)

Topic: Expert Systems Exercise

1/29 Review: EXSYS Manual (R)

Video: EXSYS Watch before class (Library Reference Desk)

Software: EXSYS: Incorporate 3 new Rules (Lab)

Topic: Demand Assessment & Forecasting

2/3 Read: pg. 43-50 (GL)

Software: EXPER*, REGRESS*

Topic: Expert Systems Applications

2/5 Read: Chapter 15 "Expert Systems from the Outside" (R); &

Read:"Developing Marketing Expert Systems: An Application to International Negotiations" by Rangaswamy, Burke, Eliashberg & Wind (R)

Read:"A Knowledge-Based System for Advertising Design" by Rangaswamy, Burke, Wind & Eliashberg (R)

Software: Negotex, Adcad

Genetic Algorithms for Product Design

2/10 Read: "Triangulation in Decision Support Systems: Algorithms for Product Design" Balakrishnan and Jacob (R)

Read: "Genetic Algorithms for Product Design" Balakrishnan and Jacob (R)

Software: GENELIN; Cruise the Internet for GA Freeware;

Artifical Intelligence for Market Segmentation

2/12 Read: "A Gentle Introduction to Genetic Algorithms" (R)

Skim: "Comparative Performance of the FSCL Neural Network and K-means algorithm for market segmentation" Balakrishnan, et al. (R)

Software: Modify GA Freeware; copy project data

Topic: Genetic Algorithms for Market Segmentation

2/19 Read: "Optimization of Control Parameters for Genetic Algorithms"

Software: GENERATOR

Topic: Genetic Algorithms Applications

2/24 Cruise the Internet and Library

Presentations by Students

Topic: New Product Analysis

2/26 Read: pg. 107-118

Software: BASS*; TRIALRPT*

Topic: Pricing Decisions

3/3 Read: Chapter 4, pg. 56-67 (GL)

Software: PRICE*; BID* (except Q3)

Topic: Advertising, Sales PromotionsTopic:

3/5 Read: pg. 75-81 (GL), Chapter 4, pg. 81-83 (GL):

Software: VIDALE*, PROMO*

Topic: Marketing Strategy

3/10 Read: Chapter 5, pg. 91-107

Software: GE*, COMPAD*

Genetic Algorithms Project

3/12 Work on Project

Software: GENERATOR

Genetic Algorithms

3/17 Project Presentations

Final Paper Due

------------------------------------------------

GRADING OF SPREADSHEET ASSIGNMENTS

Criteria

1. Answers to the Questions at the end of the Assigned Exercise in a professional looking Report

1. Indicate in your Report as to how you think the Model Should/Could be improved

1. Actual implementation of the improvement. Need to Demonstrate and write-up in the Report.

1. Show through your discussion that you understand the structure of the embedded model. (E.g., What are the assumptions; equations used, etc.)