rolling-mean
Input Stack:| minNumValues: Int | | n: Int | | expr: TimeSeriesExpr |
|
⇨ |
Output Stack: |
Compute a rolling average over a specified window of datapoints. The mean is only calculated
when there are sufficient non-NaN values in the window, making it robust against missing data.
This is useful for smoothing noisy data while maintaining data quality requirements.
Parameters
- expr: The time series expression to compute the rolling mean for
- n: Window size (number of datapoints including current value)
- minNumValues: Minimum required non-NaN values to compute the mean
Window Behavior
- Window-based: Uses a fixed number of datapoints, not time duration
- Quality threshold: Requires minimum non-NaN values before emitting a result
- Missing data handling: NaN values are ignored in calculation but count toward window size
- Consolidation impact: Each datapoint represents a longer time window when step size increases
Data Processing Example
| Input |
3,2,:rolling-mean |
| 0 |
NaN |
| 1 |
0.5 |
| -1 |
0.0 |
| NaN |
0.0 |
| NaN |
NaN |
| 0 |
NaN |
| 1 |
0.5 |
| 1 |
0.667 |
| 1 |
1 |
| 0 |
0.667 |
Examples
5-point rolling mean with minimum 3 values:
Since: 1.6
