This project is related to news coverage regarding COVID-19 around the world. Using LDA topic modeling, we have extracted topics in news about COVID- 19 over time and from different parts of the world. We aim to comprehensively analyze the correlation between the impression of the US public on pandemic from April 2020 to August 2020 together with the proportion of news reports in the United States, China and South Korea, utilizing tools such as the correlation coefficient and Granger causality. The impressions include distress about spreading viruses, being kept in isolation, stressing about financial hardship and failing economics. We use the rationality of these results to compare the reality people are being faced with within the real world.
We are given data about the proportion of news topics in different countries and the general impression of the pandemic in the US. Since there is a lot of missing and non-consecutive data, we cannot use interpolation to fill the NaN values. We decided to focus on the complete data gathered by certain countries, including China, Korea and the US’s news topics from April 4th to August 8th and public impressions from April 4th to August 8th. We use this data to compute both Spearman correlation coefficients and Granger Causality.
In the future, we hope that this data helps assess how the media affects certain parts of the world and how much more data we still need to help gauge the correlation between media and sentiment.
Presentation (PPT) presented online to classmates and professor Derry Wijaya via the Gather.Town app
Thanks to Ding Ma, Gordon Ng, and Yining Wang for working alongside me!