Setting analysis parameters for a variable type
To run an IBM® Quality Early Warning System parametric analysis, you must set the parameters. A variable type analysis is for numeric data such as temperature, volume, or height.
About this task
| Parameter name | Data type | Description and setting |
|---|---|---|
| Runtime | Time | The time point from when time is counted back for the number of points that are included in the analysis. The default value is the system time for when the analysis is run. The format must be yyyy-MM-dd HH:mm:ss. |
| Sigma | Double | The assumed standard deviation of the measurements. The value must be greater than zero. |
| Target | Double | The most desirable value for the center or mean measurement of the measured population. |
| Acceptable level | Double | The deviation from the mean of the measurements that is acceptable. Typically this level is close to the Target value and reflects the amount of movement for the population mean around the target. Where these is low process capability this level coincides with the Target, indicating that there is no movement for the population mean. |
| Unacceptable level | Double | The deviation from the mean of the measurements that is unacceptable. You must have good detection capability for this level to work well. The unacceptable level must be further the Target than the acceptable level. Maintain a degree of separation between the acceptable and unacceptable levels of 0.2*Sigma. |
| Type of control | int | If Type_of_control = 1, then the control is one-sided. Only changes up or changes down are detected. If Type_of_control = 2, then the control is two-sided: both types of deviation from the Target are detected. For example: If Type_of_Control = 1 and Accept_Level < Unaccept_Level, then only increased changes of the process mean are flagged. If Type_of_Control = 1 and Accept_Level > Unaccept_Level, then only decreased changes of the process mean down are flagged. If Type_of_Control = 2 then you specify either Accept_Level < Unaccept_Level or Accept_Level > Unaccept_Level, with the understanding that the acceptable and unacceptable levels of the two-sided procedure are positioned symmetrically around the target. |
| False alarm rate | Double | The default value is 5000. The detection procedure produces a rate of false alarms of 1 per 5000 points when the population mean is located at the LAM0. Generally, it is a good policy for the false alarm rate be at least 50X the expected window size (number of points); in this case, the false flagging probability per analysis is about 0.02. |
| Number of points | int | Total number of points in the analysis. The default value is 100. |
Procedure
- Select and select the variable that you want to set.
- Select Apply and Save.