SOCR-22

The Statistics Online Computational Resource (SOCR) analyzes “Big Data,” which are very large, heterogeneous, time-varying, multisource, and incomplete datasets that are difficult to interpret and model in meaningful ways using classical probability, statistical, or algorithmic approaches. The SOCR team designs and disseminates educational materials, web-services, and advanced data science methods and tools in probability, statistics, machine learning, and health analytics. This research team will:

(1) Enhance SOCR analysis toolbox and visualization components with an emphasis on Big Biomedical and Neuroscience Data. The toolbox will be designed to run in a web browser and enhance the visual presentation and interpretation of Big Data. The creation of the toolbox will allow many more researchers (including students) to learn about, appreciate, contribute, and apply complex analytics to their work, making Big Data much easier to turn into “big results.”

(2) Implement powerful, modern, and portable webapps (HTML5/JavaScript/Rshiny/Rmarkdown/Jupiter) that can be used to model various interesting processes, enable exploratory and quantitative data analyses, and facilitate the understanding of high-dimensional and complex information.

(3) Develop advanced AI/ML data analytics, e.g., compressive big data analytics, statistical obfuscation techniques, and Bayesian approaches to address specific biomedical, healthcare, neuroimaging-genetics, and other applications.

(4) Expand the novel Spacekime Analytics method for mathematical representation, statistical inference, and computational prediction of large longitudinal information.

More details are provided on the SOCR Research website (http://socr.umich.edu/html/SOCR_Research.html).

Meeting time and location: For MDP academic credit, MDP-SOCR R&D courses (e.g., ENGR 255/355, NURS 995, ENG 455/599) are hybrid – we will mainly meet face-to-face occasionally, and very often coordinate synchronously via SOCR Zoom-channel, and asynchronously via Cloud services. The entire SOCR team meets once each semester (in-person, with Zoom streaming), twice each month typically on Fridays or Tuesdays at 8 am – 9 am ET via SOCR-Zoom channel (distance synchronous communication), weekly in smaller project-specific team sessions, and coordination asynchronously of all progress, challenges, and interactions via G-Drive. Each sub-team arranges a convenient time to meet and work together following university guidelines. Annual, two-term enrollment commitments begin each January.

Team organization

Each SOCR sub-team is coached by an experienced student that reports to the SOCR faculty and the PI. Sub-teams are mostly focused around developing the mathematical foundations, building particular algorithms, and designing statistical approaches for addressing applications. The sub-teams are flexibly structured to promote creativity, provide opportunity for student growth, and nurture team-science. We have the following project sub-teams: SOCRAT, CBDA, DataSifter, Data Analytics, Data Science Fundamentals, Spacekime analytics, (see SOCR website). As students develop skills and build confidence, they should expect increasing responsibility on assignments with multiple parts of the SOCR Lab.