Process statistics

Calculating process statistics helps to determine whether a current process is under control. It also helps in determining the future performance of the process using the existing data. The statistics for a characteristic are calculated based on parameters such as specification limits, sigma estimation method. The calculation of process statistics happens when new results are recorded in the SRO.

The specification limits and control limits for a characteristic are obtained from a QM specification that is linked to the most recent sample for an entity. The result data from the range of samples are evaluated against the spec limits obtained from the most recent sample.

The range of samples for a characteristic is limited as follows:

• The samples are restricted to those that have the same value as the most recent sample for the columns corresponding to the ones where the filter_by_* settings for the characteristic is True. For example, if filter_by_item_id is True, only samples for the same item as the most recent sample are used to calculate these values.

• The difference between samples that were requested more recently than the time of the most recent sample and the cut-off time configured in the system attribute are used. This helps to limit the number of samples.

• The maximum number of samples is limited to the value configured in the system attribute. This helps to limit the number of samples.

The most recent sample record for a characteristic is represented as the starting sample. A range is obtained from the most recent to the oldest sample record, based on the values configured in the system attributes. The value for the process statistic attribute is calculated on the basis of the default chart type, which is configured for a characteristic or a QM specification that overrides the characteristic.

The statistical values are calculated in a stored procedure that is called whenever results are added. The calculated statistical values are used in Sample Viewer control, custom implementations, etc.

The statistical values are rounded off to their nearest decimal places configured for a characteristic before they are returned to the client. However, the intermediate statistical values such as range, sample average, and sample deviation are stored in the sample_char_link table. They are not rounded off to the nearest decimal places. The current values of the process statistics are also stored in the stats table by entity and characteristic and any filter fields.

The values configured for a characteristic for standard deviation and standard mean are not used to calculate the process statistics. These values are used only to calculate control limits. None of the statistic attributes listed in Statistical values are exposed in the SRO.