- type: stddev
  name: REQUIRED. Must be unique among the list.
  value: REQUIRED. The input of the standard deviation algo.
  pivot_column:
    OPTIONAL. Integer. If supplied, any item of `value` is threatened as
    a list, and the `pivot_column` is extracted from any item.
    For example, if pivot_column is 1, then the second column
    (zero-based) of every item of the value list is used for
    the calculation (see example for more details)
  compare_to: REQUIRED. Integer. The standard deviation to compare against.

​
Example

condition:
  - name: stddev-condition
    type: stddev
    value: "{{ steps.db-step.results }}"
    pivot_column: 2
    compare_to: 1

For this example, the output of db-step step is a list of rows from the db:

[(1, 2 ,3), (1, 4, 5), (7, 8, 9)]

The pivot_column is 2, hence the values for the stddev calculation are: 3, 5 and 9.

Next, the sttdev condition calculates the stddev:

standard deviation = sqrt(sum((x - mean)^2) / N)
mean = (3 + 5 + 9) / 3 = 5.666666666666667

And the standard deviation (sd) is:

standard deviation = sqrt(((3-5.666666666666667)^2 + (5-5.666666666666667)^2 + (9-5.666666666666667)^2) / 3)
                   = sqrt((9.555555555555557 + 0.11111111111111116 + 9.555555555555557) / 3)
                   = sqrt(6.740740740740742)
                   = 2.5961484292674155

Therefore, the standard deviation of the dataset [3, 5, 9] is approximately 2.596.

Thus, the values that are more than 1 standard deviation from the mean are 3 and 9, since they are outside the range of 5.666666666666667+-2.5961484292674155 (which is [3.0705, 8.2628]).

​
Same example without pivot_column

Notice that we used pivot_column since the output db db-step was a list of rows. If the output was just list, we could skip it.

For example, if the output of db-step was (3, 5 ,9), we could just use:

condition:
  - name: stddev-condition
    type: stddev
    value: "{{ steps.db-step.results }}"
    compare_to: 1
AssertWhat is a Function?