Recently I became interested in knowing more about my social media usage. I’ve been using some form of social media for over a decade now and I wanted see just how exactly I use it. My first thoughts were to Twitter because it’s the platform I’m most active on and represented the largest source of data.
The first question I wanted to answer is “How often do I tweet?” Turns out this question is one of the easier ones to answer. Here’s how I did it.
The first thing I did was download my entire Twitter Archive. Your Twitter archive to allows you to browse a snapshot of your Twitter information, starting with your first Tweet. You get two things in your twitter archive:
- A fancy browsable local index.html website that opens up a JSON file of your tweet information
- A simpler CSV file of all your tweet information
For the purposes of creating this data visual I really just wanted the CSV. So with the CSV in hand I set about cleaning the dataset and isolating the only the data I was interesting, the number of tweets per day. I deleted all the columns of data I wasn’t interested in (like individual TweetIds) to get down to just a list of tweets with timestamps.
The timestamps were in a
MM/DD/YYYY/00:00 format and since I was just interested in the day of a tweet so I left Excel and used Atom’s multi cursor function to quickly delete the hour/minute information. (It’s a neat little tick I learned for working with CSVs from my friend and coworker Tom VanAntwerp).
Once I was down to just the date timestamps for every tweet I did a simple Count Individual Values function back in Excel. This gave me a count of the number of duplicates of unique values which of course is the number of tweets in any given day.
One histogram later here we are:
Some quick insights on the data:
- I have tweeted on 1,365 different days.
- My median number of tweets per day is 7 which seems about right when I look at my Buffer account which is set for minimum five tweets per day.
- I was interested in the outlier days where I tweeted a ton but they were actually just batch uploads of Facebook image albums thanks to IFTT.
This is just the start of diving into my Twitter data. Now I’m onto a harder task of pulling my Twitter engagement data from the Twitter API.