In Phase II analysis you apply the principal component model that you built in Phase I to new process measurements, labeled "Phase II Data" in FigureĀ 10.1. The MVPMONITOR and MVPDIAGNOSE procedures read Phase II data from a data set that is specified with the DATA= option. You must also specify the OUTLOADINGS= data set that was produced by PROC MVPMODEL as a LOADINGS= data set. The loadings are used to compute principal component scores, residuals, and statistics for the Phase II data.
PROC MVPMONITOR can save the Phase II data and the corresponding computed values in an OUTHISTORY= data set. An OUTHISTORY= data set has the same structure as an OUT= data set that is produced by PROC MVPMODEL, and it can be read by the MVPMONITOR and MVPDIAGNOSE procedures as a HISTORY= data set, as shown in FigureĀ 10.1.
You use PROC MVPMONITOR to produce control charts of Phase II data in order to monitor the process for continued stability. Unusual variation might indicate that the process is no longer stable. charts detect unusual variation within the model hyperplane, whereas SPE charts detect unusual variation outside the model hyperplane. Unexpected SPE chart variation might indicate that the process variation has changed such that the principal component model no longer adequately represents it. You can save a summary of a control chart by using the OUTTABLE= option in an SPECHART or TSQUARECHART statement in PROC MVPMONITOR. This enables you to "replay" a chart by specifying a TABLE= input data set. You can also read a TABLE= data set that was created outside PROC MVPMONITOR to display a chart with customized control limits.
In addition to producing contribution plots of Phase II data, the MVPDIAGNOSE procedure also produces score plots. Score plots can provide insight into the nature of the variation that is represented by the components. You can compare score plots of Phase II data that are produced by PROC MVPDIAGNOSE to score plots of Phase I data that are produced by PROC MVPMODEL.