Science of data science

One of the most intriguing, yet challenging, aspects of data-intensive science is the manner in which it produces interdisciplinary collaborations. These collaborations result from the fact that the manipulation and analysis of large-scale data frequently requires scientists to depend on the expertise of statisticians, computer scientists, information scientists, and the like. 

Such collaborations create opportunities for scientists from a variety of disciplines to come together and develop new approaches and methodologies. However, these partnerships also create challenges because they mix together surprisingly different research cultures with diverse evaluation systems, publication practices, privacy and sharing norms, and other field-specific characteristics.

Start date: 3/21/2014
End date: 10/14/2015

Read More

An institution or organization hoping to encourage innovative data science among its faculty and researchers needs to provide the supporting infrastructure to overcome barriers to collaboration and realize the potential benefits of interdisciplinary collaborations.  

This project proposes to follow a four-step process to grow institutional capacity for data science initiatives at the University of Michigan. The research team will work jointly with three university programs that are already oriented toward data science and are committed to building greater capacity in that area.  

The first step of the project will initiate four seed projects within the context of the MCubed program, each of which brings together three to four researchers from diverse disciplines to collaborate on innovative data-intensive research. The second step will embed information science-oriented researchers into each of these projects. These researchers will carefully observe the barriers to collaboration and cultivate tools and techniques to overcome these barriers. 

The third step will feature a series of three workshops conducted over the course of the MCubed projects that will bring together the researchers and other university-level data science initiatives to review observations and devise approaches to overcome barriers. The final step is to translate the project’s findings into university-level capacity and data science support structures at the University of Michigan. An important component of this final step will be a university-wide symposium on data science following the completion of the MCubed projects.

By focusing on the configurations of data science teams, their projects, challenges, and support needs, this project’s findings could easily translate to other universities and be relevant in other institutional research contexts.

Grants

Seeding New Data Science Collaborations, Gordon and Betty Moore Foundation: $440,000 

 

The Gordon and Betty Moore Foundation believes in bold ideas that create enduring impact in the areas of environmental conservation, patient care and science.