rolling-min
Input Stack:
|
⇨ | Output Stack:
|
Find the minimum value within a rolling window of datapoints. This operation is useful for tracking low-water marks, establishing lower bounds for noisy data, and creating envelope functions that follow the lowest points in fluctuating signals.
Parameters¶
- expr: The time series expression to find rolling minimum for
- n: Window size (number of datapoints including current value)
Behavior¶
- Minimum selection: Returns the lowest value in the current window
- Missing data: NaN values are ignored in the calculation
- All NaN window: If all values in window are NaN, emits NaN
- Window-based: Uses fixed number of datapoints, not time duration
Alerting Example¶
Using rolling min to establish recent low baseline:
name,sps,:eq,:sum,
:dup,
5,:rolling-min
This creates both the current value and its recent minimum for comparison.
Data Processing Example¶
| Input | 3,:rolling-min |
|---|---|
| 0 | 0 |
| 1 | 0 |
| -1 | -1 |
| NaN | -1 |
| 0 | -1 |
| 1 | 0 |
| 1 | 0 |
| 1 | 1 |
| 1 | 1 |
| 0 | 0 |
Examples¶
5-point rolling minimum:
| Before | After |
 |  |
name,sps,:eq, :sum | name,sps,:eq, :sum, 5,:rolling-min |
Related Operations¶
- :rolling-max - Rolling maximum values
- :rolling-mean - Rolling average values
- :rolling-count - Count non-zero values in window
- :min - Global minimum across all values
Since: 1.6