Twitter open sources an anomaly-detection tool

2 Comments

Twitter has open sourced a tool that it uses to detect spikes in time-series data, in order to make sure it’s able to react to issues with its servers, applications or other systems that might affect the site’s performance. The company also uses the tool, which is an R package called BreakoutDetection, to track large upticks in engagement during big events such as the Super Bowl. According to a blog post detailing BreakoutDetection, Twitter created it in order to deal with the mass of data and relative commonality of anomalies associated with running large-scale systems.

Fig2_0

2 Comments

Bob Foster

They basically repackaged the ‘stl’ function with some plot methods. Twitter — keepin’ that stat-game on lock since 2015!

Data Scientist

C’mon guys, you’re supposed to be good at data science. This is not an anomaly detection tool, it is a breakout detection tool that is robust in the presence of anomalies…

Comments are closed.