Data-driven modeling of complex systems overview:

Data-driven modeling (particularly in social science applications) is a new area of research for me that I have been exploring as a postdoc. These projects involve data gathering (e.g., through the HuffPost and Twitter APIs), statistical techniques, uncertainty quantification, and the development of minimal models.

Minimal models for forecasting U.S. elections


Election forecasting is a high-stakes problem characterized by uncertainty, incomplete information, and subjective choices (e.g., related to defining likely voters). Interestingly, polling errors tend to be correlated in states that have similiar demographics, adding further complexity. With the goal of shedding light on the forecasting process and exploring state-state interactions, we are developing data-driven minimal models for forecasting U.S. elections. Our compartmental approach tracks the number of Democrat, Republican, and undecided voters in each state, and our parameters are fit to data from HuffPost and RealClearPolitics.com.

Our model forecast the 2018 senate and governor races with the same accuracy as more complex popular sources (e.g., FiveThirtyEight.com), and we are using our model to study how subjective choices on how to account for uncertainty impact forecast interpretations.

Collaborators:
Daniel F. Linder (Augusta University), Mason A. Porter (University of California, Los Angeles), Grzegorz A. Rempala (Ohio State University)

Preprints:

External forcing + Twitter response: NFL anthem protest dynamics


Many social movements spread on Twitter, and prominent examples include protests surrounding gun violence, harassment, and social injustice. While some hashtags fade out of use, others remain prominent in the Twitter conversation. In this collaborative project that started through a 2018 AMS Math Research Community, we have been downloading Twitter data and exploring retweet network structure. Because the NFL anthem protests have a weekly periodicity that adds another time scale to the problem, we plan to focus specifically on the NFL discussion on Twitter in the future. Through a cohort study of select Twitter accounts with a demonstrated interest in football, we will develop compartmental models to explore the interplay of tweet dynamics with news coverage and the football schedule ("external forcing").

Collaborators:
Heather Z. Brooks (University of California, Los Angeles), Maria R. D'Orsogna (California State University, Northridge), Punit Gandhi (MBI, Ohio State University), Sarah Iams (Harvard University), Kellen Meyers (University of Tennessee), Joseph Tien (Ohio State University)

Network construction from diary-based data


This summer REU project focused on lifting diary-based data from a college social community to a larger extended network. We constructed separate networks for home, social, and work interactions and tested how dynamic changes in these networks impact epidemic size. Simulating an influenza outbreak using an SIR (susceptible-infected-recovered) model, we found that reducing encounters at work after infection is an effective way of decreasing flu season severity.

Collaborators:
Joshua Rubin Abrams* (University of Arizona), Anne Schwartz* (Amazon), Veronica Ciocanel (MBI, Ohio State University), Björn Sandstede (Brown University)

* Asterisk denotes undergraduate students

Preprints:

Alexandria Volkening
Last updated July 14, 2019