Picking the Right Metric
blog.sigopt.com
blog.sigopt.com
Let's say I run a mobile game (multiplayer, online, players can interact). What can you do for me and what do I have to give you to do it?
There are two ways to use us as a mobile game developer.
1. For optimally tuning the machine learning models you may have for predicting user engagement, churn, etc. We wrote a blog post on this last week [1]. More information can be found on our machine learning page [2].
2. For A/B tests of various in-game parameters [3]. Instead of doing exhaustive search (or binary search, or local hill climbing, or intuition) to decide what parameters to try in your tests you can use SigOpt to optimally suggest what to do next, given the historical data already collected. Example: if you are trying to find what the best properties a specific new unit should have in your game (cost, build time, etc) we can guide you to the global optima for the metric you care about automatically and optimally.
We do all of this by wrapping cutting-edge design of experiments research behind a simple API [4] and web interface. You tell us what you have tried in the past (if anything) and we tell you what is optimal to try next.
Any other feedback is greatly appreciated! Feel free to shoot me an email at contact@sigopt.com any time!
[1]: http://blog.sigopt.com/post/111903668663/tuning-maching-lear...
[2]: https://sigopt.com/cases/machine_learning