rossgore.github.io

Website for collecting research work

Selected Research Projects

COVID-19 Case and Vaccine Forecasting

Leveraging Data To Better Understand The Human Experience

Simulation-Debugging, Validation & Verification

COVID-19 Case and Vaccine Forecasting

Developed a model at the city and county level of Virginia that gives a short-term forecast for the total number of COVID-19 cases, COVID-19 vaccines administered and supplied. Available at: https://vmasc.shinyapps.io/va-county-covid-forecast/.

Selected Media Coverage

Leveraging Data To Better Understand Humanity

A major focus of my research is leveraging data that is an artifact of the way we live to better understand human behavior. Typically, this data takes the form of social media data, anomyized GPS mobility data, and anomyized SMS messages and cell phone records. My aim is to novel insight about human behavior by quantifying and analyzing these data sources.

You Are What You Tweet

We studied the relationship among the obesity rate of urban areas and expressions of happiness, diet and physical activity on social media. We showed that areas with lower obesity rates: (1) have happier tweets and frequently discuss (2) food, particularly fruits and vegetables, and (3) physical activities of any intensity.

Selected Media Coverage

Understanding How Locals And Tourists React Differently To City Attractions

We studied how the time of day and if individuals were locals or tourists can affect the sentiment individuals express towards attractions. We showed that: (1) tourists express more positive sentiment towards attractions than locals and (2) more positive sentiment is expressed about attractions in the morning vs. the afternoon / evening.

Characterizing The Mobile Phone Use Patterns Of Refugee Hosting Provinces In Turkey

We used an anomyized mobile phone data set to understand the experience of refugees throughout Turkey based on their proximity to refugee targeted violent incidents and refugee camps. We showed that: (1) there is more negative sentiment targeted expressed towards refugees in those areas close to refugee camps and (2) the mobile phone behavior (i.e. number of calls made and text messages sent) in the wake of a violent incident for refugees and non-refugees is more similar the close the individuals are to the location of the incident. The research won the “Safety and Security Prize” in the Data For Refugees (D4R) challenge sponsored by Turk Telecom.

Mobile Data As A Public Health Decision Enabler

We used anomyized cell-phone data from Senegal provided by the Data For Development challenge and joined with with demographic data and the location of hospitals to identify those areas in the country where adding an additional hospital would most reduce the travel time during the critical initial period for patients experiencing an heart attack or stroke. The research won the “Partical Application Prize” in the Data For Development (D4D) challenge sponsored by The Orange Foundation

Modeling Human Behavior With Agent Based Models

They are diverse in their knowledge and abilities but their behavior is far from random. However, despite often making rational decisions, their behavior is also emotional. My research with agent-based models has attempted to capture these attributes, when they matter, to better understand how humans operate and how the institutions around them operate in a variety of domains.

Selected Media Coverage

Simulation Debugging, Validation And Verification

The process of developing, verifying and validating models and simulations should be straightforward. Unfortunately, following conventional development approaches can render a model design that appeared complete and robust into an incomplete, incoherent and invalid simulation during implementation. An alternative approach is for subject matter experts (SMEs) to employ formal methods to describe their models. However, formal methods are rarely used in practice due to their intimidating syntax and semantics rooted in mathematics.

To address this problem I have developed an approach to gaining insight about unexpected outputs, in some cases bugs, centered around the practice of predicate-based statistical debugging used in software engineering. This approach is realized in a standalone tool published on my Github and in an online web application. We have a small but regular user base (~25-50 users) and we are always looking to grow it.