Coronavirus Tweet Analysis Project

The full code is available on Github

Main Question: How are People Responding to Coronavirus?

1. Webscraping of Coronavirus Tweets

• Polarity & Subjectivity: We scraped tweets with the word "coronavirus" in it in US.
• Economic & Diff Countries: We scraped tweets with the word "coronavirus" in it for top 10 corona countries.
• Time period: February 11th, 2020 ~ April 21st, 2020
• Location: US (We divided US into US_east and US_west due to geolocation limitations and combined the two)
• Process: To webscrape, we used the Python3 library GetOldTweets3and the setTopTweets() to scrape only the most frequently retweeted tweets.

Webscraping Code

Below are the functions in GetOldTweets3

This is the code for webscraping the coronavirus tweets

The same process was done for us_east and then we combined the two csv files

This is what our final csv looks like. It has two columns, the Date and the Tweets

Data Engineering - Subjectivity Polarity

# Subjectivity & Polarity

In conducting our sentiment analysis we used the sentiment analysis API provided byTextBlob The polarity score is a float within the range [-1.0, 1.0]. The subjectivity is a float within the range [0.0, 1.0] where 0.0 is very objective and 1.0 is very subjective.

In our data engineering process we calculated the subjectivity and polarity of our data. A high subjectivity rate means that the tweet is highly opinionated. Negative polarities point to tweets showing negative emotion while positive tweets refer to tweets showing positive emotion. After calculating these values, we added two new columns, subjectivity and polarity to our dataset.

This is how our final csv looks like. You can find this csv file HERE

Data Engineering Code

Data Engineering -Economic terms for each Country

Overall we were able to see increases of tweets when significant events occured