Ninjas attack

Ninja2Clearly demarcated as blatant self-promotion: I'm happy to announce that an effort I've been working on for about 3 years in semi-stealth mode is now live. After doing big data work for the spooks in the government and running a team of social and computer scientists, a few of us spun out a commercial venture I dubbed Ninja Metrics (www.ninjametrics.com).* Much of this comes from constantly asking game companies for data, then getting smart people to do cool things with it. Our team has now published 80 papers on game data, which is kind of ridiculous. It was time to put that power into an engine so we don't have to work quite so hard.

Among other things, we've figured out how to automate predictive analytics. WTH is that, you ask?

Our system will automatically take in data and spit out all of the stats and metrics one would expect. DAU, ARPU, yadda yadda. That was the easy bit. The cool part is that the system actually builds models on the fly. As in it takes the raw data, cleans it, then can tell you the predictive likelihood of a player to do something next week. Spend, quit, convert, cheat, etc. It's running about 80-90% accurate.

The coolest thing we do relates to players interacting with each other but I'll save that for a second post.

If you're interested in the science of games and data analysis, I've started a blog on the topic, which can be found here. Please feel free to subscribe or link to it (this makes my marketing guy happy).

And if you'd like me to tackle anything data-related here at TN, fire away and I'll dig in. After hiding underground for three years, I'm about ready to be public again. The 10-year anniversary makes me nostalgic for the days when we were stomping around arguing that games were important.

Also, and this is a long-term goal,  it's my hope that by getting a load of game data I'll be able to anonymize it at some point and open it to the broader research community. I have seen what a great group can do with just a little and am excited to see more.

 

*Why? I like ninjas. Isn't that a good enough reason?

 

 


Comments on Ninjas attack:

ron meiners says:

Congrats Dmitri! It sounds awesome already, and I've the feeling it's the start of an evolution in sophisticated game design... Anyway, looking forward to learning more as you move forward.

Posted Sep 25, 2013 1:44:59 PM | link

Anthony Panozzo says:

The link to Ninja Metrics is not working. Looks like it is somehow relative when it should be absolute. Hope this helps!

Posted Sep 25, 2013 7:13:40 PM | link

Dmitri Williams says:

Weird! Thanks for catching, fixed. At the risk of the obvious, it's www.ninjametrics.com.

Posted Sep 25, 2013 8:03:38 PM | link

Andrew Peterson says:

Does Ninja Metrics work with any Learning Management System?

Posted Sep 26, 2013 4:39:02 PM | link

Dmitri Williams says:

We haven't, but as we look outside of gaming we're thinking about how our social value system could help in other spaces. The keys are that the experience is social, data can be collected via API or from an existing database, and that we can somehow build social graphs of the end-users. In gaming, that's easy since they play, talk and do joint sessions. In other verticals it often means we need to piggyback on to an existing graph like Facebook, which in turn necessitates a social sign in.

Feel free to ping me off-list as well as raise issues/questions/challenges here.

Posted Sep 26, 2013 6:12:57 PM | link

greglas says:

Congrats, Dmitri!

Looks awesome.

Posted Oct 3, 2013 9:36:24 PM | link

greglas says:

Congrats, Dmitri!

Looks awesome.

Posted Oct 3, 2013 9:36:25 PM | link