rolling-sum
Input Stack:| n: Int | | expr: TimeSeriesExpr |
|
⇨ |
Output Stack: |
Compute the cumulative sum of values within a rolling window of N datapoints. This operation
is useful for calculating totals over recent periods, tracking cumulative events, or smoothing
data by summing neighboring values.
Parameters
- expr: The time series expression to apply the rolling sum to
- n: Window size in number of datapoints (positive integer, includes current value)
Window Behavior
- Datapoint-based: Uses a fixed number of datapoints, not time duration
- Inclusive: Includes the current datapoint in the sum calculation
- NaN handling: NaN values are treated as 0 for summation but count toward window size
Examples
Applying a 5-datapoint rolling sum:
Data Processing Example
| Input |
3,:rolling-sum |
| 0 |
0.0 |
| 1 |
1.0 |
| -1 |
0.0 |
| NaN |
0.0 |
| NaN |
-1.0 |
| NaN |
NaN |
| 1 |
1.0 |
| 1 |
2.0 |
| 1 |
3.0 |
| 0 |
2.0 |
Consolidation Impact
Since the window is based on datapoints rather than time, the effective time coverage changes
when step size increases due to consolidation. Each datapoint represents a longer time window
when the step size is larger.
- :rolling-mean - Rolling average over datapoint windows
- :rolling-max - Rolling maximum over datapoint windows
- :rolling-min - Rolling minimum over datapoint windows
- :rolling-count - Rolling count of non-NaN values
- :integral - Cumulative sum over entire time window (not rolling)
- :sum - Total aggregation across time series
Since: 1.6
