Ken Wood

If you Tweet in the US…

Blog Post created by Ken Wood Employee on May 7, 2014

If you tweet in the US…


Do you Tweet in the US? If you do, chances are, you are doing so from an Apple device. I recently watched a “new car commercial highlighting enhanced integration through your iPhone called ‘Siri Eyes Free’” – and I know why.


I hope this article sparks an interesting debate in the HDS Community. I’m sure everyone has an opinion on this subject.


But, before I tell you how I know this, let me set this up. It’s been a while since I’ve posted anything on the HDS Community. Sometimes, I work on projects where I can share bits of information, essentially, beating around the technical bush as it were, with related topics or subject matter relating to the core of some project I or my team is working on that is destined to be released in the future. So, while these projects are a secret, I usually try and drop Easter egg-blogs to peak your interests and bring you up to speed on the subject leading up to some announcement.


Recently, since late last year, I’ve been working on a project that has been not only a secret, but that I have been consumed in. Only recently as I start documenting our findings, and closing out the project, am I starting to tell the story of the data collected, processed and visualized. These long, multi-stage data flow scenarios have been a rewarding challenge for my team and I, and now I get to start telling the stories of what, how and why. Basically, we are collecting, pre-processing, analyzing and visualizing data.


You may already know this, but for me, I sometimes drink the Koolaid being served only because I haven’t had thetime to do the research myself. The Koolaid here is “Android devices are outselling Apple”.  Sample of charts showing the decline of Apple mobile devices from a random source I picked because


of the nice pie chart.


This is not an iPhone versus Android piece. This is showing some results of data we’ve been collecting and visualizing. The chart here also doesn't show usage, just sales, which may very well be true.

Well, technically now, I have done some

research. There are several use cases and different data sets that we have worked with over these past 5 months, but the most interesting insight has come as part of Phase 3 (the final phase of this project).


This is the simplest version of this data (I’ll share additional perspectives in later posts). Take a look at the geomaps charts we generated below.




What these charts “say”, is this – “if you Tweet in the continental US, you mostly like are tweeting from an Apple device”. In this case, an Apple device is defined as an iPhone or iPad. In these charts, the Orange colored dots are the source location of tweets via Twitters latitude and longitude coordinates. Red dots are Android based devices. There are also Yellow and Green devices which represent BlackBerry and Microsoft devices (collectively), and Unknown devices, but those are almost non-existent. This does not incorporate other factors like the demographics of Apple and Android users, or the time of day that this blob sample of Twitter data was captured (though every capture period results in a similar pattern).


There are other assumptions that can be deduced from these charts like the situation of South America’s high density of Android devices versus North America’s high density of Apple devices. A similar distinction can be interpreted between northern Europe versus southern Europe. Again, same with Japan and southeast Asia. I have my hypothesis, but I won’t share them here.


So, what does twitter data look like? Here is one tweet. For a 20 character tweet, this is the JSON data that is create about that one tweet and what we collect as a stream – tons of them. You can see the actual message is tiny compared to all the data about this data. We plucked the device type source of the message if it is available, and the latitude and longitude for the location of the source tweet if it is available.JSONSample.jpg


For this particular set of geomap charts, these are the only fields we used. In other visuals (the subject of subsequent posts), we combine additional fields to processing to gain yet another perspective of the Twitter-verse.


A Use Case

The use case here might start with market research on what features a company’s next vehicle might incorporate. As in the car commercial mentioned above, this vehicle includes a system that tightly integrates with an iPhone and Siri. Why would they choose this mobile device particularly for the US? Again, study the geomap charts. You might not do this in Brazil. This would also allow a public relations campaign via social media – Twitter -to craft their message is such a way as to appeal to Apple users. Again, understanding other demographic information of today’s iPhone users.


Granted, this is a simple social media research effort and lacks a lot of additional demographic information about consumer buying tendencies, the demographics of Apple users, what type of person tweets. It also doesn’t include other social media networks like Tumblr, Instagram, FaceBook, and so on. A simple study could be performed on all of these networks and then correlated for a more accurate market landscape, but you get the idea.


I’m curious to hear your opinion and comments about this.