integral
Input Stack: |
⇨ |
Output Stack: |
Compute the running cumulative sum (integral) of a time series across the evaluation time window.
Each output datapoint represents the total accumulated value from the start of the time range
to that point. This is useful for calculating totals from rate data or estimating the area
under a curve.
Parameters
- expr: The time series expression to compute the integral of
Behavior
- Each output value is the cumulative sum:
integral[i] = sum(input[0] to input[i])
- Missing values (NaN) are treated as zeros
- Negative values will decrease the cumulative total
- The result represents the area under the input curve
Data Processing
Input |
:integral |
0 |
0 |
1 |
1 |
-1 |
0 |
NaN |
0 |
0 |
0 |
1 |
1 |
2 |
3 |
1 |
4 |
1 |
5 |
0 |
5 |
Examples
Basic integral of a constant:
Integrating rate data to get total counts:
Counter Metrics
For counter metrics, datapoints represent average
rates per second. To get meaningful totals, first convert to per-step values using
:per-step, then apply :integral
to get the cumulative count.
- :derivative - Rate of change (inverse operation)
- :per-step - Convert rates for use with integral
- :add - Single-step addition vs. cumulative sum
- :rolling-sum - Rolling accumulation (related to cumulative operations)