What do youth health, climate change, and the coronavirus have in common? The answer: they are the themes of recent data challenges for U-M faculty and students, sponsored by the Michigan Institute for Data Science (MIDAS).
A data set seeking solutions
When faculty members of U-M Family Medicine started examining the youth health and wellness dataset from MyVoice, a national text-message poll, they wondered what creative analyses the data science students could come up with for this dataset. This idea quickly became a data challenge that MIDAS coordinated along with the faculty members.
The 2021 MyVoice Data Challenge on Youth Health and Wellness attracted 17 graduate and undergraduate student teams. They used natural language processing and other data science methods to analyze the data and synthesize findings, which will help inform local and national policies in real-time. They researched which social issues were most important to American youth, the impact of COVID-19 on mental health, examined evidence of a correlation between race and concerns about COVID-19, and many more topics.
This is one of the many data challenges that MIDAS coordinated in the past two years. The increasing prevalence of data challenges reflects the need for data-driven decision making in complex scenarios.
How it works
A MIDAS data challenge typically lasts 2 to 8 weeks, involves students with diverse skill levels, and is usually co-sponsored either by external organizations or U-M faculty members. Students form teams to examine a particular dataset to answer open-ended research questions that the organizers provide or from the students themselves. External sponsors and faculty members benefit by having their data examined from many perspectives as well as accessing student talents who may be an asset to their organization as interns or full-time employees.
MIDAS organizes on-campus challenges, but U-M students seek opportunities to engage in external data challenges and have done very well internationally. In January 2021, a team of undergraduates selected by MIDAS won the Best Paper Award in the Weather and Natural Disaster Prediction category of ProjectX, a machine learning and climate change data competition. With mentoring from Dr. Sindhu Kutty (Computer Science and Engineering), the U-M team competed against peers from 22 top universities in North America and Europe. Another student team led by MIDAS faculty members Dr. Brahmajee Nallamothu and Dr. Ji Zhu won the grand prize of the American Heart Association’s COVID-19 data challenge for their project, “Population-based features and their association with coronavirus disease 2019 infection in the United States”.
Many data challenges are brought forward from industry partners who seek opportunities to work with data scientists at U-M and recruit talented students. They also have a complex business problem in need of creative solutions, and U-M data experts are a great source for innovative ideas. Data challenges can also take a variety of fun formats. In March 2021, for example, a MIDAS student team will be competing against data scientists from one of our industry partners in a week-long Data Challenge.
To learn more about how to co-sponsor a data challenge with MIDAS, please contact Kristin Burgard, MIDAS Outreach and Partnership Manager, firstname.lastname@example.org.