How hyperpartisan publishers are tapping into Twitter is a Twitter analysis and data vis story for the MSc Digital Journalism course at Goldsmiths, University of London.
I scraped over 60,000 tweets using Python (code in twitter_follow_acct.py). The scraper ran for just over 3 days, over the period of a weekend. Methodology is described in the text of the story.
View the the data in this spreadsheet. The xls file is sorted into pivot tables summarising the data, but there's also a sheet that has the raw data from the scraped tweets.
Data vis was made using Highcharts.js library. Web page built with Uikit.