Research and Development position in Analytics, Data Mining, Machine Learning.
• Data Mining, Data Analysis, Knowledge Discovery, Machine Learning, Artificial Intelligence, Neural
Networks, Decision Trees, Multilinear Regression, Logistic Regression, Expert Systems, Genetic Algorithms,
Natural Language Processing, Feature Subset Selection.
• WWW based application development using Perl, Java, CGI, DBI, Servlets, HTML, DHTML and XML.
• Development and usage of C++, STL libraries, Object Oriented Perl modules and Java libraries.
• Unix (Solaris, AIX), Linux and Microsoft Windows environments.
• Relational Databases (Oracle, MS SQL Server), Database Design, ERWin, Visio
October 2006 – February 2008
Senior Staff Scientist
BasePoint Analytics LLC, Carlsbad, CA
• Developed credit and debit card fraud models. Models are developed using feed forward neural networks.
• Developed a new method for explaining the outputs from Neural Networks for the credit card fraud
• Developed new Perl scripts for statistical data evaluation and data validation and transformation.
• Developed new C++ programs for model evaluation.
• Developed real time referral optimization techniques for real time fraud detection for credit card
companies to stop the riskiest fraud transactions at the point of sale. With the optimization techniques we
developed we maximize the amount of fraud detected with minimum number of referred or declined
• Worked on new product development for detecting credit card fraud for online transactions.
• Worked on scoring and evaluation solutions for mortgage fraud and early payment default for Mortgage
Lenders and Investment Banks.
April 2003 – September 2006
Senior Staff Scientist
Fair Isaac Corporation, San Diego, CA
• Designed and developed transaction data driven predictive models for credit and debit card fraud detection,
credit card risk analysis/prediction, predicting credit card attrition and account profitability using neural
networks, machine learning, statistical modeling, pattern recognition, and artificial intelligence.
• The main responsibilities included: analyzing and understanding large amounts of historical transactional
data to determine suitability for modeling, data clean-up and filtering, pattern identification and feature
extraction, feature (variable) selection, experimenting with different types of algorithms and modeling
techniques, analyzing performance, preparing model reports for communication with internal and external
clients, participating in pre-sales, and providing post implementation support.
• Supervised and trained junior analytic modelers.
• Developed new software tools and libraries used in model development, data analysis and conversion and
model performance analysis.
Jan 2002 – March 2003
AT&T Labs, Middletown, NJ
• Developed System and Design Requirements for Integrated Order Management (IOM) System.
• Designed and maintained the Design of IOM Oracle Database by using ERWin and Visio.
• Designed screens for IOM application which is a web based application.
• IOM is intended to provide a common user portal to AT&T Business Services’ service-specific ordering
infrastructures (Data/IP, Voice, Local Voice) to facilitate customer-concentric ordering of both standalone
and bundled services.
Jan 2000 – Dec 2001
Senior Software Designer and Software Development Team Leader
IBM Global Services, Middletown, NJ
• Software development team leader in two WWW based database application projects.
• Both applications are Object Oriented Perl CGI applications using DBI to connect to ORACLE database in
Unix (Solaris) environment.
• Developed Object Oriented Perl Modules for CGI, HTML and Database access.
• Developed an easy to use PERL XML parser that can be used for parsing or constructing XML files.
• Database access modules enable user to access the tables and fields in the database by using their names
directly that makes programming easier and shortens development time.
1998 - 1999
Lehigh University, Bethlehem, PA
• Conducted research on development of a large Internet enabled Java based database application.
• Supervised 15 undergraduate students working in the project.
• Instructed lectures on JDBC and SQL to undergraduate students working in the project.
1995 - 1998
Data Mining Research Scientist
AT&T Segmentation Scoring Platform District – Data Mining team, Piscataway, NJ
• Conducted research on customer status prediction and knowledge discovery in Segmentation and Scoring
Platform Database in collaboration with AT&T Labs researchers. Used neural networks and symbolic
machine learning techniques (C4.5 and Ripper).
• Created a system which would help marketing managers predict future customer status and extract
interesting and usable knowledge from the database. Used customer call usage data, customer status
information and demographics data for prediction and data mining.
• Developed a new technique for feature subset selection using Neural Networks.
• Developed C++ classes for making handling of very large data files easier for Segmentation and Scoring of
long distance customer data warehouse in Unix (Solaris) environment.
• Developed a multi-threaded sort program for sorting very large data files in Unix (Solaris) environment.
1993 – 1995
Lehigh University, Bethlehem, PA
• Conducted research on knowledge based engineering design and modeling. Used several C++ class
libraries and CLIPS expert system shell.
1987 - 1991
Senior Software Design Engineer
Nortel Networks R&D, Istanbul, Turkey
• Designed and developed software packages (e.g. 3-Way conference call, wake up service) for DMS
exchanges for Nortel Networks.
• Supervised five software designers and consulted thirty software designers.
• Instructed the course "Introduction to PROTEL and DMS development environment".
1991 - 1993 NATO Science scholarship from TUBITAK (The Scientific and Research Council of Turkey) for Ph.D. in
Lehigh University, Bethlehem, PA.
Ph.D. Computer Science, December 1999
Fields of specialization: Artificial Intelligence, Data Mining, Machine Learning, Decision Trees, Neural
Networks, Expert Systems, Feature Subset Selection, Discretization.
Ph.D. Thesis: "Converting A Trained Neural Network To A Decision Tree - DecText - Decision Tree
In real life applications (especially in Data Mining applications) understandability of the knowledge acquired
by a Machine Learning system is very important. DecText addresses the understandability problem of Neural
Networks by converting the knowledge acquired by the Neural Network into a classical C4.5 like Decision
Tree. Another contribution of this thesis is a new Feature Subset Selection technique. Choosing a subset of the
features will increase accuracy and reduce complexity of the acquired knowledge. Tested the new technique on
real world and artificial datasets and compared its results with existing methods. Showed that the new method
chooses good subsets by searching fewer states than the existing methods.
Developed libraries and programs in C++ and STL in Linux environment for testing the results of the research
Istanbul Technical University, Istanbul, Turkey
M.S. Control and Computer Engineering, February 1990.
Fields of specialization: Global optimization, numerical analysis techniques and optimization in control
M.S. Thesis: "A new global optimization technique and application to control system design."
Bogazici University, Istanbul, Turkey
B.S. Electrical and Electronics Engineering, July 1987.
Member, ACM SIGKDD
Boz, Olcay (2002). “Extracting Decision Trees From Trained Neural Networks”, KDD-2002 The Eighth ACM
SIGKDD International Conference on Knowledge Discovery and Data Mining , July 23 - 26, 2002
Boz, Olcay (2002). “Converting A Trained Neural Network To A Decision Tree DecText - Decision Tree Extractor”,
The 2002 International Conference on Machine Learning and Applications (ICMLA'02) June 24 - 27, 2002
Boz, Olcay (2002). “Feature Subset Selection by Using Sorted Feature Relevance”, The 2002 International Conference
on Machine Learning and Applications (ICMLA'02) June 24 - 27, 2002
Boz, Olcay (1997). “Customer Status Prediction and Knowledge Discovery in Segmentation Scoring Platform
Database” AT&T Technical Report, AT&T September 1997.
Boz, Olcay (1995). “Knowledge Integration and Rule Extraction in Neural Networks.” Technical Report, EECS, Lehigh
University. October 1995.
Decker, R., Lee, D.H., Jin, H., Boz, O. (1995). "Generic Engineering Design Using Design Scripts", 9th Conference
with Industry, Information Resources in Manufacturing, Center For Manufacturing Systems Engineering, Lehigh
University, Bethlehem, PA, May 23-24, 1995.
Tigli, O., Eksin, I., Ertem, S., Boz., O. (1994)."Global Optimization Method in Controlling a Plant", Elsevier Science
1994, Electric Power Systems Research.
Tigli, O., Eksin, I., Ertem, S., Boz. O. (1992). "Optimal Control of a Plant Using Global Optimization Method",
Modeling and Simulation, Vol 23, Part4, Proceedings of the twenty-third Annual Pittsburgh Conference, April 30 - May
1, 1992, Pittsburgh, PA.
Tigli, O., Eksin, I., Ertem, S., Boz., O. (1992). "Global Optimization Method in Controlling a Plant", Third biennial
Symposium on Industrial Electric Power Applications, New Orleans, Louisiana, 1992.
Boz, Olcay, Eksin, I., Goren, L. (1989). “A New Global Optimization Technique for Multivariable Functions and
Application to the Design of Systems”, Third National Conference in Electrical Engineering, Istanbul, Turkey, 1989.
Boz, Olcay (194). “Expert Knowledge Management System. Porting Lupes to X, Express Clips Integration and Tool
Domain Knowledge.”, Semiconductor Research Corporation (SRC)/Lehigh Annual Review, Lehigh University, PA.
Presented two posters on “Expert Knowledge Management System and Engineering Design” in the annual SRC/Lehigh
Annual Review in 1993 and 1994.