Figure 5
Time series of the three kinds of sentiment for the tweets about masks as predicted by VADER and a random forest classifier trained on the Sentiment140 dataset. We observe that the general trends are broadly similar, with roughly 40% of tweets being positive from May 2020 onwards. Four major events are marked on the time series. A corresponds to a WHO warning about a global mask shortage on February 7th, 2020, B corresponds to an announcement by the Health Secretary about high-quality masks being distributed to the NHS on March 16th, 2020, C corresponds to a government announcement that scientific advice would be followed regarding face masks on April 17th, 2020 and D corresponds to an announcement that governments would make face masks mandatory in shops in England on July 14th, 2020.
