During the last two months of the MDS program (mid April to late June each year), our students work in teams with an external capstone partner and a teaching fellow mentor to address a question facing the capstone partner’s organization using data science. This page provides an overview of the MDS capstone project. You may also be interested in the partner information, the timeline and procedures.
The Capstone Projects
Successful MDS capstone projects…
- pose an interesting and open-ended question/problem that can be addressed using data science for which data is available or obtainable.
- pose a multi-faceted question/problem, containing enough dimensions to be addressed in a multitude of ways.
- are sufficiently deep, such that a useful data product can be made in two months that makes a solid advance on the problem.
- can be split up into milestones, such that concrete progress can be made in two months.
- draw on various tools and topics the students have learned in their courses during the MDS program.
The work involved in each capstone project, completed by a group of ~4 MDS students, must include
- refining the project’s over-arching question into one that can be directly addressed using data science;
- using data science to draw useful information and recommendations from data; and
- making the information and recommendations accessible to the capstone partner through various effective means of communication (such as documentation, visuals, presentations, and reports).
What a capstone project is not:
- Setting up and/or maintaining a database.
- This type of project does not focus on drawing useful information from data.
- A pre-specified approach, such as applying a particular machine learning algorithm to a particular data set.
- This type of project is too narrow, not open-ended, and has only one or a few solutions. We expect our students to creatively come up with their own data science approaches to address the capstone partner’s over-arching question/problem (potentially with advice from the capstone partner on data science approaches).
- Data cleaning.
- Although we expect some data wrangling to be involved in the project, we ask that you provide data that is mostly clean and “ready-to-go”. If you need help with this, ASDa from the UBC Department of Statistics might be able to help for a fee.
We expect our student groups to
- communicate productively, identify sub-problems that could be worked on individually by team members, and integrate contributions of team members into a final product;
- work at least four full-time days per week on the project;
- document and present their work at UBC (using written, oral, and visual means) at various points throughout the course.
- optionally (but strongly recommended), present their final product and findings to the capstone partner’s organization.
- BC Stats: Empowering employee engagement through AI – Government of BC
- Coast Mountain Bus Company: Forecasting Transit Schedules and Congestion Areas
- Rio Tinto: Tools for Analyzing Mining Drill Data
- UBC Centre for Teaching, Learning and Technology: edXvis: Interactive Visualization of Student Engagement with edX MOOCs
- Unbounce: Unbounce Community Forum Analysis