Analytics Education

Present and future of analytics education

Panel discussion: Directors from four of the nation’s top university analytics programs share their insights and expertise.

By Kaibo Liu, Diego Klabjan, David Shmoys and Joel Sokol

Education-Learning-dotshock

Many universities have established programs or concentrations in data analytics to meet the job market demand. Image © dotshock | 123rf.com

According to the IBM Tech Trends Report in 2011, analytics has been identified as one of the four major technology trends in the 2010s. However, as predicted by McKinsey Global Institute, by 2018, the United States alone will face a shortage of more than 140,000 people with deep analytics skills, as well as a shortfall of 1.5 million managers without knowledge on how to leverage the big data techniques to make effective decisions. Consequently, many universities have been establishing master’s degree programs or concentrations in data analytics to fill the resource requirements in the current job market. While some efforts have been made in the past, there is still a lack of formal discussions and communications among these analytics programs, which poses a significant challenge for students and industries to understand the unique strengths and characteristics of different analytics programs.

This article is intended to provide some insights regarding this topic, which are digested from a featured panel discussion on “Present and Future of Analytics Programs” at the Industrial and Systems Engineering Research Conference (ISERC) that was held on May 30-June 2, 2015, in Nashville, Tenn. In particular, four panelists joined this session, including professors Joel Sokol (Georgia Institute of Technology), Diego Klabjan (Northwestern University), David Shmoys (Cornell University) and Michael Rappa (North Carolina State University), who are all the current program directors from their respective analytics programs. Professor Kaibo Liu (University of Wisconsin-Madison) chaired the session.

The panelists discussed a wide range of topics, including the effective ways of teaching analytics courses, the balanced designs of course curriculum from multidisciplines, the unique strengths and resources of different analytics programs, the expected skills for students after the analytics training program, the experience and challenges of the past and current analytics program, and future strategies for better education, improvement and sustainability of the analytics program. The goal of this article is to push the frontiers and increase the exposures of the analytics programs by establishing an interactive forum for better discussion and sharing of information among the analytics community. With such efforts, we intend to provide: 1) insights for students to decide which analytics programs to apply for, 2) information and guidelines for industrial companies to better understand the expected skills of students after the analytics program trainings, and 3) useful experience and successful stories for other universities who are also interested in establishing an analytics program in the future.

Analytics Program Overview

Education-Diego-Klabjan

Diego Klabjan

Education-Mike-Rappa

Michael Rappa

Education-Shmoys

David Shmoys

Education-Sokol

Joel Sokol

Based on the information provided from the four panelists, Table 1 shows a detailed summary of the similarities and differences among the four analytics programs featured in the panel discussion. In addition, some unique features of each analytics program are summarized as follows:

Cornell University offers a concentration in data analytics as part of the Master of Engineering degree in operations research and information engineering. The data analytics concentration is designed in the way to complement the current strong operation research core curriculum. The data analytics curriculum includes many OR courses, and also two statistical data analysis courses, one data technology course, and one marketing and pricing strategy course.

Georgia Tech offers a Master of Science degree in analytics. The curriculum of this program is an integration of the strengths from three colleges: College of Computing, College of Engineering and Scheller College of Business. The program allows students to build skills based on three tracks of specifications, including analytical tools, business analytics and computational data analytics. In addition, it also provides flexible course selections based on students’ interests (half of courses are elective).

Northwestern University offers a Master of Science degree in analytics. The program provides a highly applied and comprehensive curriculum that integrates three disciplines: data science, information technology and business. The courses are all newly designed and taught by faculty from different fields and departments of the university. This program aims to equip students with the essential techniques of extracting and communicating the value of data. During the program trainings, the students are exposed to a wide range of software tools, including R, Python, D3, SAS, Tableau and Hadoop (MapReduce, Spark, Pig and HBase).

North Carolina State University offers a Master of Science program in analytics that is run by the Institute for Advanced Analytics, a university-wide collaboration that is organized independently of college and department units. It is widely recognized as the first Master of Science in analytics program in the country. One unique feature of the program is that there is no conventional concept of courses; instead, faculty members directly determine the appropriate lectures and materials that they think useful for students. Also, the program does not intend to grade and rank students; instead, it emphasizes a balance of mixed skills and trainings, including technical, tools, teamwork, communication and problem-solving with an integrated multidisciplinary curriculum in math, statistics and computer science.

Key Tools and Software for Analytics Students

While there are many choices and differences in the tool/software offered by school analytics programs, the panelists indicated that the ability to learn a new tool is more important than learning a particular tool/software, as software selection will continuously change and evolve over time. Upon graduation from an analytics program, the students are expected to acquire some popular tools, such as R, to build a “toolbox” that they can consistently build upon. However, students need to realize that the existing “toolbox” is always going to grow in the future, and new knowledge has to be timely updated based on their working experience and requirements to ensure their success beyond college.

The panelists also suggested that it is more important to provide a wide overview of a potential toolbox in the beginning of the program and then guide students to actively teach themselves about the details of each programming language. In this way, students can not only achieve a basic understanding of the strengths and uniqueness of different tools and software, but also acquire the skills of actively learning a new tool when it becomes important. Furthermore, the panelists also highlighted the essentials of teamwork training in the analytics program, which helps students to better communicate in a future working environment with others who may use different programming languages.

Design Models of Analytics Programs

The panelists provided insights and shared useful information regarding the design models of their respective analytics programs, which should be helpful to those who are interested in establishing new data analytics programs at their universities in the future. Considering that analytics is an interdisciplinary field, the panelists suggested that it could be more effective to consider the program as a “standalone” unit rather than an “additional” element in an existing programs, such as statistics, computer science, business or industrial engineering. In this way, faculty from different departments can contribute to the analytics program with their unique expertise via course instruction and student advice.

The panelists also commented that it is more important to provide a breadth of curriculum to students rather than focusing on introducing in-depth knowledge in a particular topic in the analytics program. Students are expected to have enough prior knowledge in different fields and to have the ability to quickly learn new knowledge in their daily life going forward.

  Cornell Georgia Tech NC State Northwestern
Degree granted: M.Eng. in ORIE with a Data Analytics concentration M.S. in Analytics M.S. in Analytics M.S. in Analytics
Year started: 2004 2014 2007 2012
Location: Ithaca, NY Atlanta, GA Raleigh, NC Evanston, IL
Number of faculty: 25 58 NA 12
Program format: Full time Full or Part time Full time Full time
Time duration in months: 9 or 14 12 10 15
Curriculum: Core & Electives Core & Electives Fully Defined Fully Defined
Enrollment per year: 3~6 about 45 about 115 <=40
% Female: NA 38% 41% 45%
% International: NA 42% 15% 40%
Acceptance rate: NA 18% 13% NA
Applicant undergraduate GPA requirements: NA NA NA >=3.0
Applicant TOEFL requirements: Total >=100 Total >=100 No Total >=95
Application fee $95 $75 with $2,000 matriculation fee NA $75
Applicant GRE/GMAT requirements: Only GRE accepted GRE or GMAT accepted No GRE or GMAT accepted
Matriculated students  with prior work experience: NA 60% >50% 56%
Number of graduates: <30 45 419 61
Sponsored team project: Yes Yes Yes Yes
Internship requirement: No Yes NO Yes
Average starting salary: NA $95,000 $96,400 $103,250
Latest job placement by 90-days after graduation: 100% 100% 100% 100%
Total tuition cost for in state/out of state students: $48,900 $39,000/ $52,400 $25,000/ $42,800 $64,800
Program directors: David Shmoys
(ORIE Chair)
Kathryn Caggiano
(MEng Director)
Joel Sokol Michael Rappa Diego Klabjan

Should Analytics Programs Provide Online or a Residential Degree?

While there have been many discussions about offering online degrees, the panelists agreed that the residential degree for their analytics program is essential, at least for now. For example, according to panelist Joel Sokol, “One of the key aspects of Georgia Tech’s program is what goes on outside of the classroom. We offer special training sessions in leadership, ethics, creativity, communication, privacy, etc., as well as a required applied analytics practicum. Those are hard to offer in an online format. Some of the other aspects of the curriculum are also part ‘art’ and part science, such as data wrangling, building good modeling intuition, learning how to get projects implemented, etc., and those also might be hard to translate to an online format.”

However, the panelists also acknowledged the advantages of the online courses, such as reaching broad audience, and thus believe that certainly parts of the curriculum could be offered online in the future. These online courses will serve as a complementary role to the residential program.  

Analytics or Data Science Degree?

The panelists agreed that most of the top data science programs focus mainly on the computing and statistics pieces (some include a little O.R.), but they do not cover the business side. This is a unique difference between Master of Science programs in analytics and data science. For example, Georgia Tech’s master’s degree in analytics is interdisciplinary between engineering (statistics and operations research), computing and business, and the inclusion of the business piece is a significant differentiator for the program. The students are more likely to come out with a broader business understanding of the use and application of the technical side, and are therefore hopefully more likely to advance to significant leadership positions within organizations in the long run.

Another important difference is that an analytics program is more about understanding the tools available to provide better decision-making support and using them to guide data-driven decision-making. While analytics has certain overlap with data science, the focus on the application of big data, rather than simply the manipulation and maintenance of the data itself, is a basic distinction between the two disciplines.

Collaborations with Healthcare Informatics

Analytics is a very broad field. There are many healthcare nursing and physician informatics programs currently available in the United States. Those programs are mainly certification-based, and the curriculum is often not an analytics- or statistics-driven program. Thus, many clinical experts are willing to get more intensive training in the analytics programs. One challenging question is how to better collaborate with clinical experts and ensure the analytics program is suitable to those groups of people with a different skill or background.

The panelists suggested two solutions: One is to provide more practical/clinical-related elective courses in the curriculum, and the other is to establish more collaborations with clinical providers through healthcare-related analytics projects. For example, David Shmoys from Cornell University mentioned that the students in his program commonly conduct two analytics projects a year with providers such as the Cayuga Medical Center in Ithaca, N.Y., the Hospital for Special Surgery in New York City, and the New York Presbyterian Hospital/Weill Cornell Medical College. Diego Klabjan from Northwestern University mentioned that his analytics program indeed has bioinformatics students.

While these students are required to take technical/IT coursework, at the same time they are also flexible to choose elective courses based on their interests, e.g., genomics instead of marketing. Sokol added that his program provides many healthcare analytics electives to students and applied practicum projects in healthcare analytics to accommodate the ever-growing needs in this field. In addition, one of the program’s industry advisory board members is a director for a healthcare provider.

Kaibo Liu is an assistant professor at the department of Industrial and Systems Engineering, University of Wisconsin-Madison. His research interests are system informatics and data analytics with an emphasis on data fusion for process modeling, monitoring, diagnosis and prognostics.

Diego Klabjan is a professor at Northwestern University, Department of Industrial Engineering and Management Sciences, and a founding director of the Master of Science in Analytics program at Northwestern. He is also a founder of Opex Analytics LLC.

David Shmoys is the Laibe/Acheson Professor and director of the School of Operations Research and Information Engineering at Cornell University. He is a Fellow of INFORMS, SIAM and ACM, and a recipient of the Lanchester Prize from INFORMS.

Joel Sokol is the Fouts Family Associate Professor in Georgia Tech’s H. Milton Stewart School of Industrial and Systems Engineering. He is founding director of Georgia Tech’s interdisciplinary Master of Science in analytics degree program, served two terms as INFORMS vice president of Education, and is the recipient of Georgia Tech’s highest awards for teaching and student impact.