Ben Weber's Blog
Data Science Manager at Twitch. Studied game AI and machine learning at UC Santa Cruz. @bgweber
An overview of how Twitch uses staged rollouts to perform experiments, biases we found, and methods we use for significance testing.
Some of the events leading up to the announcement of the Blizzard and DeepMind AI collaboration.
My thoughts on the challenges DeepMind will face if applied to StarCraft
My 2015 playlist includes prior years.
An introduction to adding telemetry to a game.
A comparison of behavior trees and the ABL planning language.
Ben Weber's Comments
[Blog - 01/09/2015 - 01:32]
This is an interesting post ...
This is an interesting post Dmitri and I 'd like to hear how you distinguish predictive analytics from forecasting. In my experience, as soon as you start sharing predictive modeling with a game team they will start asking for forecasts. r n r nThe way I like to differentiate these ...
[Blog - 09/24/2014 - 03:01]
Great article Dmitri, I m ...
Great article Dmitri, I m looking forward to this ongoing series r n r nDo you have any examples of vendors/APIs that allow clients to directly query the transactional data being collected This would be useful for pivoting metrics based on custom events, such as the half-pipe tricks example you ...
[Blog - 07/29/2014 - 12:54]
Nice post. I would say ...
Nice post. I would say the 90 rule is way off, especially if the analyst role is not embedded on the development team. r n r nAnother area sometimes overlooked when setting up telemetry is planned use of data . There s often a focus on collecting as much data ...
[Blog - 07/17/2014 - 09:35]
Once again, great article on ...
Once again, great article on getting into the details of BTs. Looking forward to part two with parallel nodes and mixed strategic/tactical decision making. Here 's a blog post I wrote on Gamasutra comparing BTs to reactive planning: r n r nhttp://gamasutra.com/blogs/BenWeber/20120308/165151/ABL versus Behavior Trees.php
[Blog - 07/07/2014 - 08:48]
How long after launch do ...
How long after launch do games start to fit the Weibull distribution It 'd be interesting to see how the fit changes over time for a particular title.
[Blog - 05/22/2014 - 12:34]
One of the biggest issues ...
One of the biggest issues in meeting the demand for data scientists in games is a lack of data sets that students can use for building the necessary expertise. Sure, you can write a script to pull data from a web site such as the Steam Gauge tool, and even ...