BUSINESS MODELING AS FOUNDATION IN DEVELOPING
DATA MINING TOOLS
Case of Automobile Warranty Data
Dimitar Christozov
American University - Blagoevgrad, Blagoevgrad, Bulgaria
Dgc@aubg.bg
Abstract: Using business models is essential for successful contemporary management. A model represents
understanding about the real life domain and reflects its essential traits. Another characteristic of today’s
management is availability of huge amount of accessible and searchable data accumulated over executing
business activities. Exploring those data to increase domain understanding and to build models, which
further can be used to support decision making is the essential task of the emerging area of business
intelligence and especially data mining. The process of modeling based on business data analysis to build
decision support tools is discussed via sharing the experience acquired in a data mining project. The project
aimed at discovering factors influencing the warranty cost in automobile industry. The warranty data
analysis has to serve for solving two different tasks: (i) design of warranty policy. Warranty is an important
marketing tool, used to share the risk of failures between all customers. This share is included into the
product’s price. Also, warranty is an important advertising and promotional tool – warranty coverage
encourages the customers to purchase the product. In its both purposes, the warranty policy requires careful
analysis related to the cost, which influence pricing and overall marketing policy. (ii) increase the reliability
of cars. Warranty data contains information about the most common problems leading to failures. Warranty
analysis helps to identify priorities and directions for improving the products. Or how to improve the cars
and to reduce the warranty cost. An iterative research process of developing and exploring models to
facilitate data analysis is presented. The process includes the following phase: data collection; data research;
modeling; defining analytical procedures to expose the discovered patterns; development software tools to
support use of the analytical procedures. Interpretation of discovered patterns provides the necessary
arguments in design of software tools for regular ongoing business analysis to support decision making. The
applied methodology represents a good practice in specifying, designing and implementing components of a
data mining application.
BRIEF BIOGRAPHY
Dimitat Christozov is a Professor of Computer
Science at the American University in Bulgaria,
Blagoevgrad 2700, Bulgaria since 1993 and at the
University of Library Studies and Information
Technologies since 2002. He has more than 30 years
of research and education experience in areas as
computer science, applied statistics, information
systems. His recent interests are in the field of
business intelligence and data mining. He graduated
Mathematics from Sofia University “St. Kliment
Ohridski” in 1979. He completed his PhD thesis
“Computer Aided Evaluation of Machine
Reliability” in 1986. and DSc thesis “Quantitative
measures of the quality of informing” in 2009. In
ICTT “Informa” (1986-1993) Dr. Christozov was
involved in establishing the national information
network for technology transfer and research in the
areas of technologies assessment, integral quality
measures and information systems for quality
management. In these areas he was recognized as
one of the leading experts in Bulgaria. At the
American University in Bulgaria, he was the leading
person in curriculum development, launching and
development of the majors of Computer Science
(1993) and Information Systems (2008). At the
University of Library Studies and Information
Technologies he proposed and implemented the
major of Information Brokerage. Professor
Christozov has more than 80 publications as separate
volume, journal papers and papers in refereed
proceedings. He is a founding member of Informing
Science Institute and chair of Bulgarian Informing
5
Christozov D.
BUSINESS MODELING AS FOUNDATION IN DEVELOPING DATA MINING TOOLSCase of Automobile Warranty Data.
In Proceedings of the First International Symposium on Business Modeling and Software Design (BMSD 2011), pages 5-6
ISBN: 978-989-8425-68-3
Copyright
c
2011 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
Science Society; and founding member of the
Bulgarian Statistical Society and the Bulgarian
Telework Association.
BMSD 2011 - First International Symposium on Business Modeling and Software Design
6