You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardexpand all lines: docs/querying/functions.md
+8-3
Original file line number
Diff line number
Diff line change
@@ -380,17 +380,22 @@ do not show up in the returned vector.
380
380
Similarly, `histogram_stdvar(v instant-vector)` returns the estimated standard
381
381
variance of observations in a native histogram.
382
382
383
-
## `holt_winters()`
383
+
## `double_exponential_smoothing()`
384
384
385
385
**This function has to be enabled via the [feature flag](../feature_flags.md#experimental-promql-functions)`--enable-feature=promql-experimental-functions`.**
386
386
387
-
`holt_winters(v range-vector, sf scalar, tf scalar)` produces a smoothed value
387
+
`double_exponential_smoothing(v range-vector, sf scalar, tf scalar)` produces a smoothed value
388
388
for time series based on the range in `v`. The lower the smoothing factor `sf`,
389
389
the more importance is given to old data. The higher the trend factor `tf`, the
390
390
more trends in the data is considered. Both `sf` and `tf` must be between 0 and
391
391
1.
392
+
For additional details, refer to [NIST Engineering Statistics Handbook](https://www.itl.nist.gov/div898/handbook/pmc/section4/pmc433.htm).
393
+
In Prometheus V2 this function was called `holt_winters`. This caused confusion
394
+
since the Holt-Winters method usually refers to triple exponential smoothing.
395
+
Double exponential smoothing as implemented here is also referred to as "Holt
396
+
Linear".
392
397
393
-
`holt_winters` should only be used with gauges.
398
+
`double_exponential_smoothing` should only be used with gauges.
0 commit comments