2nd Place Winner of Fall 2020
By Patrick Hinson, Maria Parnell
Abstract: In order to mitigate the spread of COVID-19, we must better understand the connections between various factors and the advancement of the virus. Using Facebook’s data on global mobility and Johns Hopkins’s data on new confirmed cases, we can further investigate the relationship between mobility within regions of Virginia and the number of new cases in those areas. Facebook’s data contains information on both "mobility" and "stationary" scores on a county level, which was combined and lagged five days in relation to the case data in order to mimic the COVID-19 incubation period, as well as information about whether Virginia was under lockdown or reopening restrictions at the time. We sought to find a linear relationship between lockdown phases and the mobility of residents on the total new cases in the area five days later. Using an Artificial Neural Network and a Random Forest Regressor model, we were able to achieve RMSEs of 14.5 and 16.21, respectively. Though these were not fully accurate at predicting new cases, we were still able to conclude that the levels in which people move and stay in does have a significant impact on the virus’s spread.
Comments