Statistics runtime behavior

The following describes the statistics runtime behavior:

• If the input value is BAD or UNCERTAIN quality, then all calculated values quality is set to UNCERTAIN. The calculations will continue and the values quality will remain UNCERTAIN until the Reset attribute is set TRUE and the quality of the input value is GOOD.

• Upon deploy of a UserDefined instance when the InputSource is BAD, the quality of all calculating statistics is INITIALIZING.

• If AccessMode is Write (Output), the statistics will accumulate based on what you used to poke the attribute.

• If AccessMode is Read (Input) or Read/Write (InputOutput), the statistics will accumulate based on the InputSource.

• If the Attribute value quality is INITIALIZING, the quality of the calculated statistics will be set to UNCERTAIN and the calculations will continue.

  Note: Enabling statistics increases the system resources utilization due to the checkpointing of all calculated statistical values. Since a majority of these calculations are updated on every execution cycle of the engine, they are checkpointed on every execution cycle.

When calculating TimeSinceLastTransition:

• If there is no change in the input value, the TimeSinceLastTransition elapsed time value increments to show the new elapsed time value since the last value change.

• If there is a change in the input value, the TimeSinceLastTransition elapsed time value resets.

On failover:

• All calculated statistics attribute values are retained and the quality is set to UNCERTAIN.

• The calculated statistics will continue when the quality of the Attribute is GOOD or UNCERTAIN.

When the Attribute is set to NaN, all calculated statistic attributes quality is set to "UNCERTAIN" and the calculations continue based on the last value of GOOD quality.

For calculating the average, the following equation is used:

• Average = sum of SampleSize in the collection divided by the total number of SampleSize in the collection.

For calculating standard deviation, the following equation is used:

• StandardDeviation = sqrt(([sum](pow((value-Mean), 2)) / (numberOfSamples-1))