Add make-series, mv-expand, parse-kv, parse-where, project-rename #389
Conversation
| The `make-series` operator creates time series data by aggregating values over specified time bins. You use it to turn event-based data into evenly spaced intervals, which is useful for visualizing trends, comparing metrics over time, or performing anomaly detection. | ||
|
|
||
| You find this operator useful when you want to: | ||
|
|
||
| - Analyze trends in metrics such as request duration, error rates, or throughput. | ||
| - Prepare data for charting in dashboards where regular time intervals are required. | ||
| - Aggregate trace or log data into time buckets for performance monitoring or incident analysis. |
There was a problem hiding this comment.
hum no, this is describing a summarize operator. They seem similiar for sure! but different.
- summarize takes in event data and lets you use aggregations to produce time series data encoded in rows
- make-series takes in event data and lets you use aggregations to produce time-series data encoded in field arrays
you can use this query to play about with the difference between the two: https://play.axiom.co/axiom-play-qf1k/query?qid=UiH3wAA3IfX-t29ng0&relative=1
the summarize will produce individual rows for every time bucket, the make-series will encode the whole thing into a single array on the avg_metric and timestamp field
the idea with make-series is that you can turn your events into array-based time series data, manipulate that array in some way (using the series_ functions usually), then use mv-exapand to turn it back into row-based time series data.
so it's not about visualizing trends or comparing metrics, but rather it provides a way to maniuplate aggregated, time-series data.
it's difficult to wrap your head around i know, it's much easier if you just play with it.
Here's another example of using make-series to build array based time-series data, then calling a series function on that array to make it a smoothed rolling average value, then using mv-expand to turn it back into row-based time series data for graphing
https://play.axiom.co/axiom-play-qf1k/query?qid=DJ19Fz8dxAH-t29ou0&relative=1
Preview:
https://axiom-mano-apl-sep1.mintlify.app/apl/tabular-operators/make-series
https://axiom-mano-apl-sep1.mintlify.app/apl/tabular-operators/mv-expand
https://axiom-mano-apl-sep1.mintlify.app/apl/tabular-operators/parse-kv
https://axiom-mano-apl-sep1.mintlify.app/apl/tabular-operators/parse-where
https://axiom-mano-apl-sep1.mintlify.app/apl/tabular-operators/project-rename