An M.S. in Predictive Analytics from Northwestern University
I received an email recently related to my blog of a year and a half ago on the then new M.S. in Predictive Analytics program at DePaul University. The writer wanted to know if I'd heard of the just-introduced distance learning M.S. in Predictive Analytics (MSPA) curricula from Northwestern University. Though I’d called the program office a while back, I hadn't gotten specific information. So I resolved anew to see what I could find. A quick introductory note to the NW School of Continuing Studies Associate Dean of Academics Joel Shapiro put me in business.
The SCS promotes the MSPA as an ideal program for working professionals seeking to balance their studies and careers. “The online course format provides a flexible and convenient structure with engaging materials that students can work on at the times of day most convenient for them. Students also have the opportunity to interact with their faculty and peers through online discussion boards. Similar to the on-campus programs, online courses require students to utilize textbooks and other learning materials that are part of the instruction for each course.” Add to that the prestige of Northwestern and the university's commitment to analytics – another Master's program will be introduced by the McCormick School of Engineering next Fall – and you have a strategy for success.
Not even an enthusiast like Shapiro, though, could have imagined the early popularity of the MSPA: more than 175 students are taking online courses in the inaugural quarter of instruction. And at $3400 per class, the program will assuredly be profitable. While students can complete the 11 course curricula in as little as a calendar year, the average probably will be two years or more.
Most of the initial MSPA cohort have full-time jobs. Many come from IT backgrounds. Some already hold MBAs and other advanced degrees and are now looking to enhance skills in data and analytics. And a sizable number are fortunate enough to have the cost borne by employers. Criteria for admission to the selective program include undergraduate academic performance as well as meaningful experience and professional standing.
Faculty hold Ph.D. degrees and are generally employed full time in industry. Affiliating with a top-tier institution like Northwestern is a powerful draw for academics, assuring a quality stream of capable instructors. Online class size is limited to 25. MSPA leaders believe the program census will stabilize at from 300-400 students.
My questions revolving on the technical quality of the program and the distance learning experience were adroitly handled by MSPA director Tom Miller. Miller was adamant the program would more than live up to the Northwestern pedigree and deliver a rigorous and rewarding curricula. He warns prospective students that classes are challenging and will consume a lot of their away-from-work time.
The curricula consist of 11 quarter courses divided into content areas of business, IT and statistical science. The business courses focus on analytics in the corporate context and include emphases on general management, strategy and organizational alignment. A required leadership course “builds from the basic premise that leadership is learned, and looks at the theory and practice of leadership at the individual and organizational level.”
The IT content includes classes in relational database design and implementation using Oracle, and data warehousing/data mining with Postgres and Weka. Though programming experience is not required of matriculants, basic knowledge of Java, Python or other language would certainly be beneficial, as would exposure to database concepts and SQL.
Core statistics and modeling starts with classic probability and statistical methods that include distributions, estimation and hypothesis testing and uses IBM-SPSS for computer exercises. Second and third PA courses revolve on linear regression, the general linear model and econometrics extensions such as time series and simultaneous equation models. SAS and JMP are the primary statistical tools for these classes. An elective advanced modeling course covers machine learning, Bayesian, spatial, simulation, visualization and mathematical programming techniques using the R Project for Statistical Computing. This course promises to be very challenging, but rewarding to those willing to invest the considerable required time and effort. Specific electives on modeling techniques for marketing, finance, web and text round out the analytics offerings.
Unlike other advanced analytics programs, the MSPA is agnostic towards statistical technology, exposing students to proprietary IBM-SPSS and SAS/JMP as well as open source Weka and R. This contrasts to the M.S. in Analytics at North Carolina State University, which focuses on SAS.
For an Information Management slide show on universities leading the way in BI and analytics education -- including DePaul and Northwestern -- click here.
Students close out the program with a capstone project that implements an analytics case study converging business, IT and modeling, or a thesis that addresses research into modeling techniques, comparative performance, predictive modeling competitions, etc.
I came away from my discussions with Shapiro and Miller bullish on the prospects for the MSPA. The core foci of business, technology and statistics sounds as much to me like data science as it does predictive analytics – and that's a good thing. The analytics sequence appears to be comprehensive and rigorous. And I like the business and practical emphases of the project management, leadership and capstone courses.
The relational database and data warehousing courses seem like a solid introduction to the technology side of analytics. I suspect the MSPA will beef up its IT focus in time, adding classes in data/statistical programming and big data handling. At that point the program might be one of the first online data science curricula in the market.
The MSPA may challenge the traditional business educational sequence of technical undergrad to MBA. I've been thinking for some time that it makes sense for many to reverse that path and learn business first before emphasizing data science. How about instead an undergrad business or engineering focus followed by several years of job experience and the MSPA?