Monday, June 27, 2016

Instagram In My Hood (1 Year of EVERYONE's Lexington, KY Instagram Posts)



PREFACE: This page contains LARGE-SCALE dataviz! It will NOT work on your phone! Walk or run to a desktop/laptop/tablet computer to view the dataviz properly formatted!

A little over a year ago I came across an amazing IFTTT (If This Then That) recipe for "Instagram in My Hood" and I thought "well I'm going to look at this just for the name..." and what I found was fantastic! It was an IFTTT recipe for cataloging ALL the geo-tagged Instagram posts within a region!

UNfortunately, Instagram's usage policy changed and now those location-based IFTTT recipes will no longer function due to changes in their API. BOOOO! =/

When I checked to see what all I'd gathered since turning it on I found it had run from March 26th, 2015 until June 1st, 2016 so a GOOD chunk of data! Of course I would have like to have run it multiple years to see if trends change or if predictors held true but alas, that's not the world we live in. Instead I can show you when and where certain people talk about certain things in Lexington!

Let's talk for a second about what this data IS:

  • PUBLIC Instagram posts
  • GEO-LOCATED posts
  • Limited to WITHIN New Circle Road in Lexington, KY (this was approximately the limit on the area I could cover with the IFTTT Instagram API call).

What the data is NOT:
  • PRIVATE Instagram posts
  • NON GEO-LOCATED posts
Interestingly enough if you choose to geo-tag an Instagram post that makes it public regardless of settings (essentially because you are "tagging" a place). 

First let's look at posts over time and by hour-of-day and day of week. Please note that the days where there are only 6-10 posts are ones where Instagram and IFTTT had some technical glitches. Also notice that if you're interested in a particular hashtag or word you can search the text content of the Instagram posts to look for frequency with the search bar on the right of this viz screen.


You can see that (as you would expect) Friday, Saturday, Sunday are the largest post days-of-the-week but I thought Thursday (because of Thursday Night Live) might actually be the next highest day-of-week. Surprisingly the next highest is actually Tuesday for some reason! I haven't done a deeper dive into the data to figure out why yet. If anyone has any suggestions let me know!

I realized that I could figure out the average posts-per day for a place but I realized that there were some places that had tons of posts per day (I'm looking at you Wild Fig! ;-) ) but I decided to scale the size of the dot on the following image to the number of posts per day and then use the count of distinct users to help bring a "pop" to the places where there are actually large numbers of different users talking about/from.


The next thing I looked at is WHO is posting and where do particular users post from the most?

I know this is a little messy but given the number of users I wanted some color variation (highlighting didn't seem to work as well without it). You can enter a username or select from the list of names below sorted by most frequent posters. If you mouse-over the name or the bar representing their number of posts it will show you a highlight on the map of all the places that particular user posts from around Lexington.





Finally I replicated some of the functionality of Instagram's search by doing a text search as well as adding mapped locations where that thing is mentioned. Below you can see "Beer" as the search term and you'll notice it coordinates to bars but more specifically breweries in Lexington! Imagine if you could do this globally with Instagram and you could find the most talked about bars in a town!


As always if you have any questions about this dataviz or any other please feel free to hit me up on twitter @wjking0

And also this is totally how I feel when I spend the majority of my birthday writing up dataviz blog posts, playing video games, and eating donuts from the awesome North Lime Coffee and Donuts. =D
<------

No comments:

Post a Comment