PhD, Columbia University
BA, Harvard University
Dr. Eugene Pinsky first joined the faculty of Boston University in 1986, after earning his doctorate at Columbia University. He was an assistant professor of computer science in the College of Arts & Sciences until 1993, when he left BU and gained extensive industry experience designing computational methods to analyze and monitor market risk at multiple trading and investment firms, including Bright Trading, F-Squared Investments, and Harvard Management Company. He has also applied his expertise in data mining, predictive analytics, and machine learning to video advertising at Tremor Video. Dr. Pinsky’s areas of expertise include pattern recognition, clustering, regression, prediction, factor models, support vector machines, and other machine learning algorithms and methods for data analysis; multi-dimensional statistical data analysis of large time-series data; data mining and predictive analytics to uncover patterns, correlations, and trends; algorithmic trading, pricing models, financial modeling, risk, and portfolio analysis; Python, R, C/C++, VBA, Weka, MATLAB, and MySQL; Big Data technologies and visualization (Hadoop/Hive, AWS, Tableau); design and implementation of software tools for quantitative analysis; and professional curriculum and course development.
- Performance analysis
- Data Science
- Machine Learning
- MET CS 521 – Information Structures with Python
- MET CS 677 – Data Science with Python
- MET CS 795 – Directed Study
- MET CS 810 – Master’s Thesis in Computer Science
Recent Research Reports
Pinsky, E., and R. Sunitsky. “Some Results from The Microstructure Analysis of the FX Limit Order Book.” White paper, Trading Cross Connects, March 2010.
Pinsky, E., and R. Sunitsky. “High Frequency Trading and Market Inefficiencies: A Statistical Physics Viewpoint.” White paper, Trading Cross Connects, October 2009.
Conway, A., E. Pinsky, and S. Tridandapani. “Efficient Decomposition Methods for the Analysis of Multi-facility Blocking Models.” Journal of the ACM 41, no. 4 (1994): 648–675.
Pinsky, E., P. Stirpe. “Performance Analysis of an Asynchronous Multirate Crossabr with Bursty Traffic.” Computer Communication Review 22, no. 4 (October 1992): 150–160.
Pinsky, E., A. Conway, and W. Liu. “Blocking Formulae for the Engset Model.” IEEE Transactions on Communications 42, no. 6 (1994): 2213–2214.
Pinsky, E., and A. Conway. “Mean-Value Analysis of Multifacility Blocking Models with State-Dependent Arrivals.” Journal of Performance Evaluation vol. 24 (1996): 303–309.
“A Simple Approximation for the Erlang Loss Function.” Journal of Performance Evaluation 15, no. 3 (September 1992): 155–161.
Pinsky, E., and A. Conway. “Computational Algorithms for Blocking Probabilities in Circuit-Switched Networks.” Annals of Operations Research 35, no. 1 (February 1992): 31–41.
Pinsky, E., and A. Conway. “Exact Computation of Blocking Probabilities in State-Dependent Multi-Facility Blocking Models.” Proceedings of the IFIP WG 7.3 International Conference on Performance of Distributed Systems and Integrated Communication Networks. North-Holland Publishing Co., Amsterdam, The Netherlands (1992): 383–392.
Pinsky, E., and Y. Yemini. “Asymptotic Analysis of Some Packet Radio Systems.” IEEE Journal on Selected Areas in Communications 4, no. 6 (September 1986): 938–946.
Pinsky, E. and A. Conway. “Exact Computation of Blocking Probabilities in State-Dependent Multi-Facility Blocking Models.” In Performance Evaluation of Distributed Systems and Integrated Communication Networks, edited by T. Hasegawa, et al. (North-Holland Publishing Co., Amsterdam, The Netherlands, 1992): 383–393.
Pinsky, E. and W. Wang. “Application of the Partition Function in Revenue-Oriented Performance Analysis for Network Management.” In Network Management and Control, edited by A. Kerschenbaum, et al. (Plenum Publishing, New York, 1990): 339–349.
Pinsky, E. and Y. Yemini. “The Canonical Approximation in Performance Analysis.” In Computer Networking and Performance Evaluation, edited by T. Hasegawa (North-Holland Publishing Co., Amsterdam, The Netherlands, 1986): 3.3.1–3.3.13.
Pinsky, E., M. Sidi, and Y. Yemini, “The Canonical Approximation in the Performance Analysis of Packet Radio Networks.” In Current Issues in Distributed Computing and Communications, edited by Y.Yemini (Computer Science Press, 1986): 140–162.
Pinsky, E., and Y. Yemini. “A Statistical Mechanics of Some Interconnected Networks.” In Proceedings of IFIP Performance ’84, Paris, France (North-Holland Publishing Co., Amsterdam, The Netherlands, 1984): 147–158.
Conway, A., and E. Pinsky. “Stochastic Modeling and Analysis of Wavelength Division Multiplexing (WDM) Lightwave Networks.” In Proceedings of the IEEE Infocom ’94, Toronto, Canada (June 1994): 560–568.
Pinsky, E., and P. Stirpe. “Performance Analysis of High-Speed Asynchronous Circuit- Switching Communication Networks.” In Proceedings of the 4th IEEE Workshop on Computer Aided Design, Analysis and Modeling of Communication Networks and Links, Quebec, Canada (September 1992).
Conway, A., and E. Pinsky. “A Decomposition Algorithm for the Exact Analysis of Circuit-Switched Networks.” In Proceedings of the IEEE Infocom ’92, Florence, Italy (May 1992): 996–1003.
Pinsky, E., and P. Stirpe. “Modeling and Analysis of Hot Spots in an Asynchronous NxN Crossabr Switch.” In Proceedings of the 1991 Annual Conference on Parallel Processing, Penn State University (1991): I.546–I.549.
Pinsky, E., and P. Stirpe. “Performance Analysis of an Asynchronous Multi-Rate Crossbar Network.” In Proceedings of the 1991 International Symposium on Communications, Taiwan (1991).
Pinsky, E., and P. Stirpe. “The Performance Analysis of an Asynchronous Non-Blocking Network Switch.” In Proceedings of the Singapore International Conference on Networks, Singapore (September 1991): 7–12.
Hsu, M., E. Pinsky, and W. Wang. “Modeling Hot Spots in Database Systems.” Extended Abstract in Proceedings of the 10th ACM SIGACT-SGMOD-SIGART Symposium on Principles of Database Systems, Denver, Colorado (1991).
Conway, A., and E. Pinsky. “Performance Analysis of Sharing Policies for Broadband Networks.” In Proceedings of the 7th International Teletraffic Congress, Morristown, New Jersey (October 1990): 11.4.1–11.4.8.
Ho, K. Y., E. Pinsky, and W. Wang. “On Improving the Efficiency of Cellular Communication Systems.” Extended abstract in Proceedings of the ISMM International Conference on Industrial, Vehicular and Space Applications of Microcomputers, New York (1990).
Pinsky, E., and W. Wang. “A Revenue-Oriented Performance Analysis of Resource Sharing in Distributed Systems.” Extended abstract in Proceedings of the ISMM International Conference on Parallel and Distributed Computing and Systems, New York (1990).
Binney, C., E. Pinsky, P. Stirpe, and W. Wang. “An Overview of Ensemble: A Software Tool for revenue-Oriented Performance Analysis of Large-Scale Circuit-Switched Networks.” In Proceedings of the 3rd IEEE Internation Workshop on Computer-Aided Modeling, Analysis and Design of Communication Links and Networks, Torino, Italy (September 1990).
Litvak, E., and E. Pinsky. “Efficient Computational Methods for Estimating Some Performance Measures in Large-Scale Communication Systems.” In Proceedings of the ORSA Conference on Operations Research in Telecommunications, Boca Raton, Florida (March 1990): 8A.1–8A.3
Ho, K. Y., E. Pinsky, and W. Wang. “Performance Analysis fo Some Channel Access Schemes in Cellular Communication Systems.” In the Proceedings of the IEEE Inforocom, San Francisco, California (June 1990): 603–610.
Pinsky, E., and W. Wang. “Computing Some Performance Bounds in Cellular Mobile Communication Systems.” In Proceedings of the International Conference on Computer Communications, New Dehli, India (December 1990): PAM 3.1–.3.6.
Binney, C., E. Pinsky, and W. Wang. “Design with Ensemble: A Tool for Performance Analysis of Circuit-Switched Networks.” In Proceedings of the SBT/IEEE International Telecommunication Symposium, Rio de Janeiro, Brazil (September 1990): 14.1.1–14.1.15.
“Applying Statistical Physics to Performance Analysis of Large-Scale Computing Systems.” Extended abstract in Proceedings of the ACM Computer Science Conference, Washington, D.C. (1990).
What is your area of expertise?
My research area is in performance analysis, data science, and machine learning. Since completing my PhD in computer science at Columbia University, I have been teaching throughout my career within both academia (Boston University, MIT) and industry. For almost twenty years, I worked in data science and machine learning areas developing applications that utilize machine learning to solve real-life practical problems, primarily in computational finance and computational advertising.
Data science and machine learning require an understanding of many interdisciplinary concepts. The practice involves mining large amounts of data and applying machine learning methods to design new algorithms to solve important practical problems. As more and more tasks become automated, machine learning has an increasing role in our life. Thus, it is important that students are taught the practical elements of data science and machine learning.
What courses do you teach at MET? What “real-life” exercises do you bring to class?
At the present time, I teach Information Structures with Python (MET CS 521) and a new course, Data Science with Python (MET CS 677). In all of my courses, I try to present many simple examples with a strong emphasis on visualization. In addition to regular homework and exams, students choose a project and present it on the last day of class.
How do you see the curriculum evolving to stay current with industry trends?
Traditionally, machine learning and data science courses have been often limited to the arenas of computer science and engineering. Such courses focus on computer science and assume students have a background in advanced math and science courses. At present, there is a growing need for students to learn and use data science and machine learning in real-life applications. The challenge, therefore, is to develop courses that are application driven and can be utilized by students with diverse backgrounds.
I am very interested in helping to develop such application-driven courses that use data mining and machine learning techniques, especially in computational finance and computational advertising. As an example, consider computational advertising. Skills in this area are in increasing demand as more advertising, commerce, and delivery of services are online. In computational advertising, we try to solve the problem of finding the best advertisement for a given target. The science behind solving this problem is very complex and requires different areas of expertise such as data mining, recommendation systems, optimization, and forecasting. As data science and machine learning algorithms are increasingly employed in advertising and marketing, it becomes essential to have a basic understanding behind this science.
I strongly believe that there is a tremendous opportunity to collaborate with my colleagues at MET and with other faculty members across the university to help develop new and extend existing courses to incorporate opportunities provided by data mining and machine learning.
What advice do you have for new students?
“Data science and machine learning require an understanding of many interdisciplinary concepts, based on classical results from mathematics and statistics. I would encourage students to study these fundamental sciences and look for solutions that are simple and amenable to intuitive interpretations. As data scientists, you should have an open mind. Negative results are just as important as positive results.”