v0.2.0

Notable changes

• Introduce a new optional anomaly_strategy label and generalize the framework (dashboards, alerts, recording rule names) to support multiple algorithms. This label can be used to select an anomaly detection algorithm on a per-metric basis.
• Introduce new robust strategy, based on the median and MAD, designed to be less affected by outliers and to detect anomalies in non-normally distributed data.
• The previous algorithm is still the default and has been named adaptive.
• Consolidates alerts and recording rules in a single file. This will help manage algorithms easier.
• In addition to per-strategy dashboards, a new generic dashboard has been introduced to make it easy to do side-by-side comparisons.
• Make label anomaly_type optional. It was initially introduced to allow setting per-type thresholds, but it isn't currently used outside of the sparse filter for the requests type.

What's Changed

• Enable support for multiple algorithms and introduce robust strategy by @jcreixell in #14

Full Changelog: v0.1.1...v0.2.0