Iterating on data collection practices for Pokemon Go

July 11, 2016

As I’m sure you can imagine (especially considering previous viz work I’ve done) the past couple of days have consisted of a lot of Pokemon Go for me.

Me walking around all weekend.

And like any good data nerd, a lot of my long poke-walks have been accompanied by thoughts about how to collect data on it. Apparently, I’m not the only one!

Okay, my fellow @Tableau #datarockstars. Please tell you are creating your own #PokemonGo data set as you play?


— Sean Miller (@kcmillersean) July 11, 2016

Go. When thinking through what the requirements should be I came up with this list:

I chose to go with an app I’ve used before for Quantified Self purposes, Nicholas Felton’s Reporter app for iPhone. The app has some really handy features like automatically geo-tagging reports, as well as adding step, weather, and photo data. You can totally customize what the questions the report asks and what kinds of answers you can give.

Adding a report

Initially, I set up questions to input which pokemon I caught and what their CP was. And, since it was just about time for my lunch break, I figured gong on a little walk and testing out my data collection couldn’t hurt. I’m glad I did because testing out my data collection process in the field helped me iterate on it and figure out what questions mattered.

All the pokemon I caught on my lunch break

As I walked, I hit a couple of Pokestops with lure modules on them. I was able to catch quite a few pokemon around them. I realized, that looking at the data, you would see a cluster of catches in this time period and might wonder why there were so many. So, I decided to add a question for if there was a lure module close by. At the same time, I realized that incense would have the similar effect, so I added a question for that, too.

I lure my pokemon like I do my interns, with lots of dranks and hella noms.

I continued my walk and hit another Pokestop where a Slowpoke was hanging out. I caught him and apparently his slowness spread to my phone, because upon catching him, my game froze. I’d estimate that around 40% of the time, the game freezes on me after I catch a pokemon. I’d like to be able to know an accurate percentage for that number. So, I added another question for if it froze or not while catching the pokemon.

“Wait….so did I actually catch that or nah?”

Going out in the real world and testing my data collection process helped me iterate on it and improve it. I was able to catch missing data earlier and it’ll lead to a more accurate dataset. The Reporter app makes it pretty easy to add new questions, so this whole experience really verified that it’s a good tool for the job. Unfortunately, Reporter is only for iOS, so if you have a suggestion for an app Android users could use, I’d love to hear it!

Now go out there and catch em all!

July 12, 2016 @ 5:23 am

Glad to see that I’m not the only one thinking of this! I’ll have to check this App out.

Harun Osmanovic
July 15, 2016 @ 2:49 am

Anyone knows of Android alternatives for these apps?


Leave a Reply

Your email address will not be published. Required fields are marked *