The Data Science Clinic at the University of Chicago partners with social impact organizations, government, industry and research groups to leverage data science research and technology to address pressing scientific, social, environmental and industry challenges. The Clinic also provides students with exposure to real-world projects and problems that transcend the conventional classroom experience including:

• • working with imperfect datasets, applying models and algorithms to real-world data, and navigating security and privacy issues

• • communicating results to a diverse set of stakeholders and translating information into actionable insights, policy briefs and software prototypes

The Clinic is an experiential project-based course where students work in teams as data scientists with real-world clients under the supervision of instructors. Students are tasked with producing key deliverables, such as data analysis, open source software, as well as final client presentations, and reports.

Clinic projects are generally scoped to run for two full quarters (10-12 weeks per quarter). Our ideal is to keep the student teams consistent across the two quarters and replace any students not continuing in a second quarter as needed for the success of the project. The four students on each  team are drawn from either undergraduate or graduate programs. Each student works between 10 to 15 hours a week. We employ SCRUM and AGILE methodology to give them a data science project management framework. The goal is to empower student growth and ownership of the project deliverables.

Each team has a weekly hour long meeting with their assigned mentor and submits a weekly progress report the night before the meeting. Mentors are drawn from research staff, postdoctoral fellows and the faculty, subject to availability, interest and needs of the project. The mentor provides intellectual guidance, direct feedback to students and serves as a sounding board for both challenges and direction. The mentors provide support and guidance on any gaps in data science knowledge by providing literature and resources.

In addition to the weekly meeting, the students meet with each other three times a week for a 15-minute standup. The goal of these meetings is to reinforce SCRUM methods and encourage students to communicate the status of their work with each other.

The students receive additional technical support from graduate TAs to ensure technical and coding challenges are supported. The Data Science Clinic Director serves as a resource for students in weekly office hours and as needed. Students and client mentors are encouraged to set up regular meetings. The meeting cadence with the external client varies for each project with some opting for weekly meetings while others meet only a few times a quarter. The expectations are that the client mentors provide domain expertise, feedback on timeline, and goal setting. It is an added strength if the client mentors also are providing additional data science mentorship, but it is not required.

There are a number of barriers to running a successful clinic course.

• Having a critical source of external and internal partners is key to having enough projects.  Building partnerships, sourcing and scoping projects are covered elsewhere in this playbook. One approach that would lower the barrier is to have multiple teams work with the same external partner on the same project or with a different research question using the same data set.
• Sourcing enough mentors from among faculty and postdocs that have data science knowledge to be an effective guide for students during these projects is another key challenge.  To address this one needs to work with chairs and deans to ensure there is an incentive structure for faculty and postdocs to engage as mentors including funding, official recognition of service, teaching credit to name a few examples.
• Another challenge is ensuring sufficient student preparation to succeed in a real world data science project. This can be addressed by maintaining a continuous improvement mindset on the entire data science major and allow for gaps and opportunities to be communicated across the curriculum.

There are a number of barriers to running a successful clinic course.

• Having a critical source of external and internal partners is key to having enough projects.  Building partnerships, sourcing and scoping projects are covered elsewhere in this playbook. One approach that would lower the barrier is to have multiple teams work with the same external partner on the same project or with a different research question using the same data set.
• Sourcing enough mentors from among faculty and postdocs that have data science knowledge to be an effective guide for students during these projects is another key challenge.  To address this one needs to work with chairs and deans to ensure there is an incentive structure for faculty and postdocs to engage as mentors including funding, official recognition of service, teaching credit to name a few examples.
• Another challenge is ensuring sufficient student preparation to succeed in a real world data science project. This can be addressed by maintaining a continuous improvement mindset on the entire data science major and allow for gaps and opportunities to be communicated across the curriculum.

The Clinic course requires a high level of administrative and technical staff support and engagement from faculty. From sourcing external partners and project scoping (example included in resources) to finding mentors with the appropriate data science experience or teaching assistants to help with code review, successfully running a Clinic is a daunting endeavor. That said there are now a number of different models for capstone and clinic that exist across the academy. If adding an experiential learning course to your program makes sense, discussing the opportunities and barriers with your chair is a great next step.

Scoping a project that fits the student skill level and course timeline is key to success. It can take a few weeks to get students up to speed on a new domain. It is common for clinic projects to take months to advance with new teams taking over in a new quarter or semester.  An external project partner that is engaged and interested in participating in a student learning experience is important. Students sometimes want to spend weeks perfecting a milestone that is already good enough. Stick to deadlines and keep the project moving forward in order to meet project deliverables. 

Providing a framework and training for new clinic mentors is mission critical. Mentors provide general project oversight and are a technical resource for the team. Mentor responsibilities include:

Project Leadership

• Guide overall project direction and objectives
• Ensure timely completion of client deliverables and documentation

Technical Guidance

• Shape methodological approach
• Advise on technical implementation

Teaching

• Provide resources for learning specific topics or technologies
• Work with TAs to guide student learning