Description |
Transforms each time series sample to the percentage of missing values according to the time scale up to it in the series (including itself). The count starts from the first actual example in the selected time range. |
Input |
Time Series – Single time series or a multiple time series. |
Parameters |
None |
Output |
Transformed Time Series - multiple |
Available in |
Dashboards |
USAGE EXAMPLE
Use the PercentMissingValue function when there are several gaps in the time series at a certain time scale. The function computes the percentage of missing values for each metric in the expression, so you can know how much of the metric is missing.
For example, the metric below obviously has gaps in the daily time scale over the last 3 months:
But how much is missing?
The last point shows that over the last 3 months, 66.2651% of the days were missing. So the metric is sparse at the daily time scale.
At the weekly time scale (over the last three months) it ends at 8.33% missing full weeks:
This can help also in setting up the time scale of an alert by looking how sparse the metric is.