Dataset of discourses about COVID-19 and financial markets from Twitter

In this data article, a collection of 11,625,887 tweets on the topic of the COVID-19 pandemic are provided. The data from Twitter were collected through Twitter API from January 2020 to June 2020. In addition, we also provided subsets of tweets containing discourses on both COVID-19 and financial topics. In order to facilitate the research on sentiment analysis, the Sentiment140 dataset containing 1,600,000 tweets that were annotated as positive or negative sentiment was also provided (Go et al., 2009) We used Term Frequency-Inverse Document Frequency (TF-IDF) algorithm to transform documents to numeric vectors and used logistic regression classifier to train and predict sentiments of tweets. These datasets may garner interest from data science, economists, social science, natural language processing, epidemiology, and public health groups.