Written by Carrie McDonald, MDI Journalism Intern
Led by MDI Fellow Dr. Helge Marahrens, the Massive Data Institute (MDI) hosted its first MDI Data Workshop of the Spring 2024 term from January 22nd to 23rd.
These sessions aim to introduce faculty, students, and staff to innovative data analysis methods, programming paradigms, and emerging technologies. This semester, MDI’s workshops are centered on non-traditional data analysis methods, with future workshops on temporal, spatial, and image analysis.
“We call them non-traditional methods because they may deviate from what you might encounter if you take an Intro to Statistics class, for example,” Marahrens said. “We want to give people methods that allow them to deal with the data that they find in the real world, rather than perfectly behaved data you would find in a classroom.”
“These types of data — network, time, space, image — are actually incredibly ubiquitous. If you go into industry, you will encounter them,” Marahrens added.
In this first workshop, entitled “Network Methods for an Interconnected World,” Marahrens focused on how to use network methods to analyze particular types of data structures with interdependencies between interconnected nodes. A departure from the conventional example of using independent trial data, this approach allows researchers to specifically study interdependencies within data. For example, as a sociologist, Marahrens uses network methods to study samples of people who are dependent on one another, such as friendship networks.
MDI Scholar Holt Cochran (MS-DSPP ’25) attended the workshop to learn more about methods he had never encountered before.
“Part of being an MDI Scholar is that I want to learn as many technical skills and hone those skills as best I can during my time with the Massive Data Institute,” Cochran said. “I have not worked with networks but I am interested in networks and this was a great introduction to that.”
Another attendee, Geunjeong Yu, a graduate student at the McCourt School of Public Policy, said that he enjoyed the workshop because it was “clear and detailed” and beneficial for his career.
“[The information] will help me to do my job better as a government official who analyzes the economic situation in Korea,” Yu said.
The upcoming workshops this semester will further explore the significance of examining interdependencies within datasets. MDI Fellow Dr. Nathan Wycoff will lead the next workshop, entitled “Timely Data Analysis: Modeling and Exploiting Temporal Correlation,” on February 26-27 at 4:00 pm – 5:30 pm.
Wycoff will lead participants through the statistical methods used most frequently by forecasters, including basic time series models such as ARIMA models and the latest machine learning tools based on neural networks. The session will also incorporate a conversation about the potentials as well as the constraints associated with forecasting. All Georgetown staff, students, and faculty members can register for this workshop here.
Later in the semester, MDI Fellows will delve into spatial and image analysis methods. More information on all of MDI’s upcoming Spring 2024 workshops is available here.