Exploring Twitter

My twitter followers are mostly strangers. As such, I wanted to know more about them. In particular, I wanted to learn what interests the people who follow me, and how does that differ what interests me personally?

Gathering Data

I started with Klout. Not the famous numeric score, but rather the influencer information that's available via the API. I decided to measure my own expressed interests by counting to see how many people I followed were designated as influential on each topic.

I setup developer accounts on Twitter and Klout, and got the desired info with the following script:

#!/usr/bin/env ruby require 'klout' require 'rubygems' require 'twitter' Twitter.configure do |config| config.consumer_key = config.consumer_secret = config.oauth_token = config.oauth_token_secret = end Klout.api_key = topics = {} Twitter.friend_ids('kevindway').each do |f_id| begin klout_id = Klout::Identity.find_by_twitter_id(f_id) user = Klout::User.new(klout_id['id']) user.topics.each do |x| topic = x['displayName'] topics[topic].nil? ? topics[topic] = 1 : topics[topic] += 1 end rescue Exception => e next end end topics.map { |k,v| puts "#{k},#{v}" }

I dropped the output of that script into Excel, and saw the following mostly unsurprising snapshot of my interests.

Business 158 Technology 97 Entrepreneurship 69 Social Media 54 Venture Capital 48 Finance 25 Science & Technology 22 Marketing 21 Economics 19 Education 18 Philadelphia 17 Product Management 16 Journalism 16 Markets 16 Books 12 Software 12 Music 12 Television 11 Movies 11 Videos 10 Money 10 Government 10 Facebook 10 Investing 10 Comedy 9 Angel Investing 8 Silicon Valley 8 Boston 7 NASA 7 Design 7 Management 6 Advertising 6 Internet Startups 6 User Experience Design 6 Creativity 6 Astronomy 6

I then repeated the procedure substituting followers for friends, and dropped that into Excel as well.

Looking For Differences

My first thought was to use Excel's conditional formatting to identify topics that were in one interest list but not the other.

All Green
All Green

This wasn't particularly informative. Virtually every topic is in both lists of interest. I decided to narrow that down by considering only the top 50 topics, instead of the 400+ topics that I had found.

Some Red
Some Red

That's a bit more useful.

Lessons Learned

I learned that my Twitter followers probably won't appreciate if I nerd out Science or Astronomy.

I also learned that my followers are quite diverse and that this diversity expresses itself in nearly every imaginable dimension, down to musical taste. (A Soulja Boy expert follows me?)

I was a tad disappointed that the disparities weren't more surprising or actionable, but it was still a worthwhile exploration.