Jordan Myerowitz completed an interdisciplinary Masters of Analytics and Data Science at the University of New Hampshire, ’[he authors asked her a series of questions related to her experiences in the applied project course which was a requirement of her program.
Tell us about the project that you worked on
A team of four, including myself, were assigned to a large healthcare company within the New England area for one of our program practicums. The company initially wanted to predict potential provider utilization throughout the year and then attempt to load-balance their providers so that they were 85% utilized. They also wanted to see how a provider’s panel affected their utilization and determine if providers were seeing patients outside of their panel on a regular basis. Finally, they wanted an operational dashboard to assist with scheduling their providers.
We went into the project with much optimism and great expectations; however, the delivery of data was extremely delayed from the initial data transfer date. Issues arose regarding Health Insurance Portability and Accountability Act (HIPAA) compliance and legal issues between both the company and university’s legal teams. Eventually, we received our data (after several iterations). Due to the nature of the data and the delays, our project scope was altered. We created a new utilization metric that measured the productivity of a provider given their scheduled appointments and hours working. This helped explain providers with “lower utilizations” still feeling busy and overwhelmed with patients, as a provider was often responsible for two patients at the same time. We were able to compare this new metric with the company’s existing metrics with a Tableau Dashboard that measured performance across company location, provider type, provider level, and individual providers across different quarters. Additionally, we provided a timeline of the number of appointments a provider was scheduled for historically. Ultimately, the company was pleased with our work and “found our project to be the most useful out of all the projects they have worked with in the past”, particularly given the reduced timeline.
What were the most valuable aspects of the project to your graduate education?
Although it was the most frustrating aspect of the project, the delays in receiving the data and the hunt and requests for more data was the most valuable aspect of the project. Classroom work, while difficult, presented us with a nice clean pre- processed dataset for in class exercises and for homework. This project more closely mirrored real-world issues surrounding the need for data and the legal hoops you have to go through in order to get it.
Sometimes after jumping through the hoops, the data is still not quite what is expected. The world of data is messy, and you’re going to get dirty playing with it.
Was there anything that you were hoping to get from the experience that you did not?
Other projects within the program were able to deploy different modeling algorithms in relation to their data. This was something that I wish we were able to do with our project but was not doable given the delays previously mentioned. Given the many factors that go into patient scheduling, it may not have been modellable with the data that was available at the time.
How much interaction did you have with the project sponsor?
Initially interaction with our project sponsor was frequent - we met in person for our first meetings and emailed after. However, our data transfer was delayed, and followup emails entailed alternative timelines and legalese to receive it, in addition to asking about other potential resources. After receiving the initial data transfer, we set up a virtual meeting to address our questions and concerns with the data. We received additional data shortly after. With the start of the global pandemic, interaction slowed down further as they are a healthcare company and were busy trying to understand what COVID-19 meant for them. We emailed on at least a biweekly basis to address questions with the data (including requests for additional data tables). Before the culmination of our project, we presented our findings to them to see if they fell in line with what they observed as an organization, and if there were any lingering questions or curiosities that they wished for us to address.
What did the project sponsor do particularly well? What did the project sponsor do particularly poorly?
The project sponsor provided us with a breadth of interesting data that was easy to access and connect to once we had it. They were agreeable and easy to communicate with. We also found that they provided useful feedback for us to continue the project in the appropriate direction. Conversely, our project sponsor gave us data to work with in an extremely delayed fashion, and they did not give us the necessary data to answer their questions. Whether it was a test or oversight, we had to ask for more data 2—3 times in order to address their business problem. This severely delayed the start of the project and led to a stressful finish of the project.
Can you briefly explain the dynamics of the team that you were assigned to?
Team dynamics were very good. I know our professors tried to design the teams in such a way to promote a high quality, finished product, and they succeeded with our team. We were all agreeable personalities and did not attempt to shirk work. We bonded over the delayed delivery of data from our company and other delays as well. One member of the team was older and had industry experience with analytics. He sought to not be the leader but was always the voice of wisdom and was wonderful at creating and styling presentations. Although he was not the most technical team member, he was very hard-working and did not shy away from coding. The other teammates were agreeable and hard-working as well. My older teammate and I often took turns leading the project and my other two teammates provided input as well. I do not consider myself an excellent coder, but I found myself coding a lot of the project as well as creating the Tableau dashboards. All in all, every team member was committed to finishing the project and helped in various ways, even if the task was new for him or her.
What suggestions would you give to students taking an analytics project/ capstone course?
Don’t be afraid to ask questions. Do your research ahead of time to see if you can answer your own questions but be sure to double-check definitions and data dictionaries with your project company. If possible, start early and constantly iterate on your finished project. You’ll run into dead ends, but that is part of the process.
Do not be afraid to try something new. It might not work for your given project in the end, but you would have learned something new and added a new skill.
Be sure to incorporate everyone into the team. If it seems like someone is not doing the tasks he or she needs to do, address it immediately before it gets worse and nip it in the bud. Not doing so will only create resentment and more problems further down the road.
What suggestions would you give to corporate sponsors in terested in working with universities?
Due to the sensitivity of data and who has access to it, please be sure to have the infrastructure in place for students to have access to the data they need to address your business problem. Whether that be the form of HIPAA authorizations, NDA disclosures, or prepping laptops and secured virtual environment - you’ll get the most out of the students if they have ample time to work on that project.
Our Summary Checklist for Working with Master’s Students
Master’s programs are different from undergraduate and doctoral programs. These students are more independent, almost always focused on the job market, and only spend half (or less) the amount of time in their program (12-24 months) relative to undergraduate and doctoral students. They are less likely to be engaged in research. All master’s-level programs have an integrated applied project requirement - which is the primary way that organizations engage with students at the master’s-level. Organizations seeking university collaborations will frequently start at the master’s- level. Our summary of the primary considerations for these collaborations include:
/ Start with the program director. Being an analytics or data science master’s-level program director is almost always a full-time position. Organizations interested in sponsoring an analytics project course should start by contacting the program director rather than a faculty member, a department chair or a dean.
/ Master’s-level analytical projects should be in the “goldilocks zone” - they have more definition than an open-ended research question but have fewer parameters than an undergraduate engagement. Students need to have the opportunity to consider a wide range of options within their portfolio of skills to get the most out of the experience.
/ The most successful master’s-level projects have a committed team from the organization who are available for questions and meet regularly with the students. Again, the “goldilocks zone” for meeting cadence is likely once or twice a month.
/ Remember that graduate students are not consultants, they are students and this is a learning engagement. Insights and innovation will hopefully be a byproduct.
/ Faculty will likely play less of a role in master’s-level projects than they do undergraduate internships or doctoral research initiatives.
/ While “data science” programs are more common at the undergraduate and doctoral levels, master’s programs are commonly titled “analytics”, or “business analytics”. However, focus less on the title of the program and more on what they actually teach. Ask to see the curriculum map.
/ Where sponsorship of an undergraduate capstone course may or may not come with an expectation of funding, most (but not all) master’s programs will expect to have the project course funded. This funding expectation may range from about $5,000- $15,000 per semester.
/ Use the master’s-level project sponsorship to inquire about other options for collaboration including undergraduate internships, doctoral-level research labs, and continuing education options for current employees.
- 1. Reid Wilson (2017). Census: More Americans have college degrees than ever before https://thehill.com/homenews/state-watch/326995-census-more-americans-have- college-degrees-than-ever-before Accessed August 1, 2020.
- 2. National Center for Education Statistics https://nces.ed.gov/ Accessed July 20, 2020.
- 3. Burtchworks Data Science/Predictive Analytics Salary Study https://www.burtchworks. com/big-data-analyst-salary/big-data-career-tips/the-burtch-works-study/ Accessed August 2, 2020.
- 4. Burtchworks Data Science/Predictive Analytics Salary Study https://www.burtchworks. com/big-data-analyst-salary/big-data-career-tips/the-burtch-works-study/ Accessed August 2, 2020.
- 5. Institute for Advanced Analytics, https://analytics.ncsu.edu. Accessed August 6,
- 6. Melissa R. Bowers, Jeffrey D. Camm, Goutam Chakraborty (2018) The Evolution of Analytics and Implications for Industry and Academic Programs. INFORMS Journal on Applied Analytics 48(6):487-499. https://doi.org/10.1287/inte.2018.0955
- 7. As an alumna of Georgia Tech, I should clarify that the Georgia Institute of Technology is an “Institute” and not a “University”. Please do not rescind my membership in the Alumni Association.
- 8. Georgia Institute of Technology, https://www.gatech.edu/academics/degrees/masters/ analytics-ms Accessed August 8, 2020.