Mapping Twitter Sentiments Towards Covid

Value-by-alpha map of sentiments with twitter representativeness level

View Full Image

This bivariate map displays Twitter-based sentiments towards COVID-19 pandemic as well as their representativeness level. The dataset is based on 738,581 geotagged English Tweets in contiguous U.S. from March 14 to June 11, collected using 74 keywords relevant to COVID-19 pandemics. The sentiments of tweet content are generated using VADER package. At county level, the average sentiment of individual tweet user is calculated in a positive-negative scale, while the representativeness level is denoted by the proportion of twitter user in total population (from American Community Survey 2014-2018). Inspired by the Value-by-Alpha map (Roth et al., 2010), we use the alpha / brightness value to illustrate the representativeness level, based on which the red color displays the positive sentiment, and the blue displays negative.

Click to learn more about value-by-alpha map.