Views from the Ground

Dr. Jeremiah Johnson, Co-Director for the B.S. Data Science Program at the University of New Hampshire, has worked with undergraduate students for over 10 years. Below, he shares his perspectives on the unique issues and opportunities for engaging undergraduate data science programs in either capstone courses or internships.

First and foremost, in order to get the most out of engagement with undergraduate populations, it is important to consider why the students are interested in the engagement (beyond the program requirement).

The best outcomes tend to occur when both the student's and the company's interests align.

Just like their graduate student counterparts (see Chapter 4), undergraduate students are primarily seeking “real-world” experience, with an eye toward improving their prospects on the job market postgraduation. While some students may have an interest in a master’s or doctoral degree, most are planning on entering the job market immediately after graduation. The student will want to build a portfolio of work, and to begin establishing their network of contacts. Given these objectives, a manager seeking to engage with an undergraduate population should ask the following questions:

■ Will this engagement provide the student with practical, “real-world” experience?

■ Will this engagement provide the student with a concrete contribution to their portfolio of work?

■ Will this engagement enable the student to begin developing their professional network?

■ Is the company interested in recruiting talented students at the bachelor level?

Even if the answer to the last question is “no”, there may still be benefit to the company in engaging with undergraduate students - a student who has a successful internship experience at a company that only hires for data science roles at the master’s or PhD level may then pursue an advanced degree with an aspiration toward a permanent position at the company. A middle way that can benefit both the company and the university is to hire post-baccalaureates contingent on completion of the requisite advanced degree, perhaps with company support provided for the student as they complete the degree. Think of this like developing a “farm team” for your major league operation.

While both capstone courses and internships are high-impact educational experiences, there are significant differences which need to be appreciated to ensure success. In particular, internship work typically takes place at the company, under the primary supervision of a manager. Capstone project work, on the other hand, typically is done at the student’s university, under the guidance and primary supervision of a faculty mentor (see Table 3-2). Critically, both should start with a meaningful, non-trivial project of some relevance to the organization.

For a full-time undergraduate student, an internship or capstone project is likely to be but one of three or four courses that the student is taking. This means that often the scope of work that an undergraduate can undertake will be less than that able to be undertaken by a graduate student over a comparable timeframe. For instance, an undergraduate student completing a for-credit hour internship will

Table 3.2 Distinctions Between Internships and Capstones



Year of Program

Typically 3rd, but can be taken from the 2nd year

Final year


On company site

On campus


Often summer, but can occur any time during the academic year

Typically during the fall or spring semester

Primary Mentor

Company manager

University faculty member

Provides College/ University Credit

Typically, but not always



If credit hours are earned for the internship, it is less likely to be paid. Credit hours are a form of "currency"


Collection of Students

Typically completed as an individual student working with a supervisor

Frequently completed as a class with other students working under a faculty mentor. The class may include students from multiple disciplines working in teams

typically be required to complete at least 150 hours of work during the course of the internship. These hours may need to be verified by an internship coordinator at the university. If the internship responsibilities are not expected to be completed over a standard 15-week semester, then the student should be expected to work ten hours per week on average, and the project to which the student is assigned should be chosen with that level of commitment in mind.

It is reasonable to expect that when you engage with an undergraduate population via an internship or capstone project, the level of difficulty or complexity of the proposed project will need to be substantially reduced from that expected of graduate students. This is true because a graduate student in data science has already completed a bachelor’s degree, and regardless of their undergraduate field of study has therefore gone through four (or more) years of an educational program that, at a minimum, grants them a certain level of maturity that may not yet be present in an undergraduate student. This will manifest, for example, in ways such as improved time management skills, communication skills, and overall comfort working in a professional environment. However, it can be said with some certainty that by the time an undergraduate student is ready for an internship or capstone project, the student has spent a substantial amount of time in their program (two years at a minimum) establishing a strong technical foundation, and this may enable them to exceed expectations. For example, students in our Bachelor of Science program in Analytics & Data Science are required to complete both an internship and a capstone project for the degree. This internship usually takes place over the summer between the student’s third and fourth years of study. At this point in the program, the student will have taken mathematics courses including applied linear algebra and calculus-based statistics. They will have learned to program in multiple programming languages, including Python, R, SAS, and C++, and they will be familiar with SQL and relational databases. Finally, they will have completed two specialized, project-based applied data science courses that integrate their technical skillset into a cohesive data analysis pipeline. This prepares them to handle a large swath of typical analytics tasks from the first day that they step into the workplace.

Given all of this, where might they come up short? Well, the average student will be very new to streaming and unstructured data. In our program, they would not have been exposed to some of the more sophisticated modeling techniques commonly used in some contexts, such as deep learning or gradient boosting. They likely will not yet have been exposed to specialized tools and techniques for working with large, unstructured datasets, and streaming data.

< Prev   CONTENTS   Source   Next >