An open source reverse-engineering of the Twee-Q project algorithm. Not because I disagree with their goals, but because I'm curious how design decisions affect the final result. This program contains 2 scripts:

• analyze_gender.rb tries to guess the genders of your retweets
• analyze_retweets.rb dumps some analysis of the timeline.

To use this, you need to run these steps:

1. On twitter.com, request your tweet archive
2. Download the tweet archive and extract it. Move the tweets directory here.
3. Run ruby analyze_gender.rb. This will dump a file tweets/retweeted_users.csv that is the result of its attempts to guess the genders of accounts you have retweeted.
4. Hand-edit the retweeted_users.csv file to correct gender guesses if you want to.
5. Then run ruby analyze_retweets.rb and it'll dump some CSV files

Example Files

If you want, I have attached the output from my own history in the example_files directory. There are three files:

• expanding.csv - when the window is expanded outwards to be larger and larger
• sliding.csv - when the 100-tweet window slides over the timeline
• sample.csv - building random samples of 100 tweets from across the over timeline