This article describes the calculations that can be used when rendering your metrics in Dashboards or Explorer.
When displaying certain metrics in a Dashboard Widget or Explorer, you can choose what calculation is used to aggregate results from multiple test runs.
The Average (arithmetic mean) is calculated by default. For example, if you configure a Number Tile widget to display Test Time (ms) for the past 24 hours, the value displayed by default would be the average of all Test Time readings from that time period. In the case of a Time Series graph, data points are aggregated based on the selected Time Interval. So if you select a Time Interval of 10 minutes, all readings within each 10 minute interval will be aggregated using the selected calculation.
You have the option to change this calculation for any metric that displays the calculation in its Metric Pill.
In the above example, the DNS (ms) metric is being added to a dashboard widget. It displays the default calculation (Avg), so by clicking on the pill, you can access the list of alternative calculations.
Calculation Options | |
|---|---|
| Average (default) | The simple arithmetic mean. Calculated by adding all of the values in the set, then dividing by the number of values. |
| 99 Percentile | The value at a particular rank, or the lowest metric value that is above a given percentage of values in the set. For example, if the 85th Percentile for a test's Response Time (ms) yesterday was 22.5ms, that means that 85% of Response Time measurements from yesterday were below 22.5ms. |
| 95 Percentile | |
| 85 Percentile | |
| 75 Percentile | |
| 25 Percentile | |
| Standard Deviation | Indicates the extent of deviation, or variance, for a group as a whole. The higher this value, the more widely the data set varies around its average. |
| IQR (Interquartile Range) | Represents how widespread the data is by dividing the ordered dataset into four quartiles (subsets with equal numbers of data points in them) and measuring the spread between the lowest and highest quartiles. IQR is calculated as follows:
|
| Median | The "middle" value of an ordered set. For this set of values - (2,4,5,9,10,10,13) - the median would be 9, as there are three lower values and three higher values. For sets with an even number of values, the middle two values are averaged to find the Median. |
| Geometric Mean | Calculated by multiplying all of the values together and then taking their Nth root, where N is the number of values. The Geometric Mean is typically preferred over the arithmetic mean (average) for datasets that have a few extreme outliers, or that vary over orders of magnitude. |
| Geometric Std. Deviation | Similar to the Standard Deviation in that it represents the extent of deviation or variance in the data set. Geometric Std. Deviation is preferred for data sets that call for the Geometric Mean. |
| Minimum | The lowest value in the set. |
| Maximum | The highest value in the set. |