MetricValue

  • Metric values represent counts that can be either exact or estimated, including an expected value and variance.

  • The JSON representation includes 'value' and 'variance' fields, both in int64 format.

  • 'value' indicates the expected metric value, while 'variance' quantifies the uncertainty around the value, being 0 for exact counts.

  • Variance enables margin of error calculation, allowing users to understand the potential range of the true metric value.

A metric value, with an expected value and a variance; represents a count that may be either exact or estimated (for example, when sampled).

JSON representation 
{
  "value": string,
  "variance": string
}
Fields
value

string (int64 format)

The expected value of the metric.
variance

string (int64 format)

The variance (for example, square of the standard deviation) of the metric value. If value is exact, variance is 0. Can be used to calculate margin of error as a percentage of value, using the following formula, where Z is the standard constant that depends on the preferred size of the confidence interval (for example, for 90% confidence interval, use Z = 1.645):

marginOfError = 100 * Z * sqrt(variance) / value