The fastest growing data science / big data profiles on Twitter

This article is co-authored with Dr. Livan Alonso Sarduy. The two charts below are based on a number of Twitter followers and their growth rate, after filtering out irrelevant or fake followers. Our next step is to identify top data science Twitter accounts, that are themselves followed by many other top data science Twitter accounts: in our opinion, this is one of the best metrics to measure real popularity.  

For some, the number of Twitter followers is a must to make a real impact. Tweets that are re-tweeted can indirectly generate as much as 30% of the traffic a website can get. Here you can find the number of followers that some data science influencers are getting per day.

Note: Twitter users with <1k followers were excluded from both graphs
Here is a list of users that have >50k followers which do not appear in the graph (* two of them have lost followers)
Twitteruser    Followers  New Followers
  • "@digiphile"     "208009" "67"
  • "@BernardMarr"   "77936"  "12"
  • "@JDavidMorris"  "60689"  "-1"
  • "@avinash"       "138334" "49"
  • "@NetworkWorld"  "57611"  "28"
  • "@dhawaldamania" "53097"  "-7"

Interesting: some data scientists with a high number of followers havea a slow follower growth. The following graph shows the relation between percentages of new followers/followers vs number of followers.

@drsanders and others do not appear in the graph because they have <1k followers

To attract more Twitter followers:

  • Build a network of people interested in your content.
  • Be positive and eloquent.
  • Spread informational content and share / re-tweet links from your followers (especially followers that are just a tiny bit more or less popular than you)
  • Don't talk about you.
  • Do not abuse hashtags, but hashtag people or companies that are in your article (they will re-tweet your article if you say good things about them)
  • It may sound familiar what someone once said: “If you don’t have anything nice to tweet, don’t tweet at all”.
  • Identify (relevant) people with very few followers (< 50), who like or retweet some of your tweets; then follow them. Once some other people decide on following them, they'll find that you are one of their few followers, and might decide to follow you as well (of course, as long as your account is relevant). This creates a little bit of a viral effect for you.
  • There are very smart ways to advertise on Twitter to quickly grow your list of followers, attracting many relevant followers, with profiles similar to yours or similar to hundreds of pre-specified profiles.

How was the data collected?

The data was collected using ad-hoc python code that reads a large input list of data science Twitter users (twitterlist.csv) and extracts information from Twitter using an API. Tthis code creates a new file with the collected information. 
The collected data was visualized using an R code. The time period corresponds to the last 10 days (specifically from 2014/07/04-2014.07/14), so the total number of new followers was divided by 10 (average follower growth per day). We will start running this code (daily or weekly) in our DSC server on Amazon EC2 (running since 07/12/2014). Also, we will make it an interactive App (maybe accessible via an API) and attach a category to each color (to classify these Twitter accounts).

Suggested Reading

Views: 10967


You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by John Barnshaw on February 19, 2019 at 4:13pm

Hi Vincent,

Can you please explain the difference from a 'logistic regression' and a 'logic regression'?  I am very familiar with the former but have never heard of the latter. 

Also, if the only difference is that the latter has independent variables binned or collapsed together in addition to the dependent variable, can you explain why anyone would ever do this?  It seems like a tremendous waste to force variation into such a dichotomized fashion.

Comment by Vincent Granville on July 17, 2014 at 7:42am

Why accounts such as @freakanomics (566,000 followers, but following 0), @FiveThirtyEight (266,000 followers, owned by ESPN), or @flowingdata are never mentioned in data science circles? I guess they are perceived as mainstream (rather than technical) and/or commercial. Probably few data scientists follow them (I probably don't, though I haven't checked).

© 2021   TechTarget, Inc.   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service