Observations made from data of followers of a few tweeters

by plainspeak


This post is in continuation of the previous two posts(https://plainspeak.wordpress.com/2012/07/16/statistics-of-a-few-twitter-handles/ , https://plainspeak.wordpress.com/2012/07/19/statistics-of-the-followers-of-a-few-tweeterscontd/)

The data collected about the followers of the tweeters is visualized using graphs(automatically generated from data using R) here. -> Graphs. The graphs are ordered in such a way that one can compare a graph of one tweeter with the corresponding graphs of all the other tweeters.The observations made from the graphs are listed below. Appreciate any feedback. The scripts used to do all the gathering of data and conversion of data to graphs can be found here -> http://github.com/rakeshbabugr/perl-twitter-harvest-data

The handles analysed are:

amishra77
auldtimer
bdutt
DilliDurAst
offstumped
PM0India
prasannavishy
realitycheckind
sagarikaghose
swamy39
swaraj_india

First we use the information about the location of the followers to come to a few inferences.

  • Significantly less number of users in cities like Chennai, Hyderabad and Kolkata as compared to other cities like Bangalore, Delhi and Mumbai. Number of followers in Chennai for a given tweeter is on an average 1/2 that of Bangalore. Same holds good for Hyderabad. Kolkata has on an average even less number of followers for a given tweeter when compared to Chennai and Hyderabad.
  • Locations in Pakistan, Kashmir and Bangladesh figuring in Top-10 locations of their followers
    • DilliDurAst – Pakistan,Karachi,Kashmir
  • Locations in Pakistan, Kashmir and Bangladesh figuring in Top-20 locations of their followers
    • DilliDurAst – Lahore,Islamabad,Srinagar
    • bdutt – Pakistan,Srinagar
    • swaraj_india – Bangladesh
  • Locations in Pakistan, Kashmir and Bangladesh figuring in Top-30 locations of their followers
    • DilliDurAst – J&K
    • amishra77 – Lahore
    • bdutt – Lahore
    • Sagarika Ghose – Srinagar, Dhaka
    • swaraj_india – Pakistan
  • Each tweeter tends to have more percentage of followers in a particular location than others(usually native place,current location)
    • DilliDurAst – Delhi
    • bdutt – Delhi
    • sagarikaghose – Kolkata
    • swamy39 – Chennai
    • auldtimer,prasannavishy,realitycheckind – Bangalore
  • Right wing tweeters tend to have a greater percentage of followers in Bangalore than the other tweeters. For instance, more number of followers of swamy39 are from Bangalore than Chennai! I have visualized the % of followers  in different cities for the tweeters -> Rplots

Now, we use the information about the followers count of the followers of the tweeters to infer stuff.

  • The % of followers with followers_count <= 10 for each tweeter is around 12%. However, 50% of bdutt’s followers have followers_count <=10. For swamy39, that figure is 42%. For PM0India it is 30%. For sagarikaghose it is 25%
  • The % of followers with followers_count <= 100 for each tweeter is around 45%. However, 90% of bdutt’s followers have followers_count <=100. For swamy39, that figure is 80%. For PM0India it is 75%. For sagarikaghose it is 72%

Next, we infer stuff from the number of tweets the followers have made in their lifetime.

  • On an average for a tweeter, the percentage of followers who have tweeted less than 10 times is 10-15%. However,50% of followers of bdutt have tweeted less than 10 times. For swamy39 the figure is  35%. PM0India – 27%. sagarikaghose – 28%
  • On an average for a tweeter, the percentage of followers who have tweeted less than 1000 times is 60%. However,90% of followers of bdutt have tweeted less than 1000 times. For swamy39 the figure is  85%. PM0India – 75%. sagarikaghose – 83%

Next, we infer stuff from the date of last tweet of the followers of a tweeter.

  • On an average for a tweeter, the percentage of followers who have tweeted in the last week is around 40%.  However, only 10% of followers of bdutt have tweeted in the last week. For swamy39 the figure is 25%,  sagarikaghose – 22%

Next, we infer stuff from what words occur most in the bios of the followers of tweeters.

  • All the tweeters have words like India, love, student occurring in the top-10 most occurring words in the bio’s of their followers.
  • Top-10 words occurring in the bio’s of DilliDurAst’s followers include : journalist, writer, politics
  • Top-10 words occurring in the bio’s of PM0India’s followers include : music,engineer,fan
  • Top-10 words occurring in the bio’s of  right-wingers include : hindu, proud, politics, nationalist, news,engineer (Perhaps the most important insight)
  • Top-10 words occurring in the bio’s of bdutt’s followers include : music,engineer,working,simple(Note word ‘politics’ not there)
  • Top-10 words occurring in the bio’s of sagarikaghose’s followers include : music,engineer,writer(Note word ‘politics’ not there)
  • Top-10 words occurring in the bio’s of swamy39’s followers include : engineer,music,working(Note word ‘politics’ not there)

Next, we infer stuff from the timezones of the followers of tweeters.

  • All tweeters have an average of 20% of followers(only those whose time-zone is visible are counted(which is around 50-75% of total followers)) from US timezone.
Advertisements