Improvements to Error-Rate Alerting

In January 2017, Mux introduced error-rate alerts in their analytics service and aimed to calculate adaptive thresholds for these alerts. However, they realized that many of the generated alerts were not significant or actionable, leading to alert fatigue among users. To address this issue, Mux trained a binary classifier using historical alert incident details to identify important alert conditions. This feature has significantly reduced the volume of alerts and increased the visibility of crucial alerts. Users are encouraged to try the alerts feature in the Mux analytics service and provide feedback for further improvements.