Examining the Principal Component Contributions

Specialized Control Charts

Examining the Principal Component Contributions

You can use the star options in the SHEWHART procedure to superimpose points on the chart with stars whose vertices represent standardized values of the squares of the three principal components used to determine T²_i.

   symbol value=none;
   proc shewhart table=prin;
      xchart tsquare*sample /
         starvertices  = (comp1 comp2 comp3)
         startype      = wedge
         cstars        = yellow
         cframe        = vigb
         cinfill       = vlib
         cconnect      = yellow
         starlegend    = none
         starlabel     = first
         staroutradius = 4
         npanelpos     = 14
         xsymbol       = mu
         nolegend ;
   run;

The chart is displayed in Figure 49.32. In situations where the principal components have a physical interpretation, the star chart can be a helpful diagnostic for determining the relative contributions of the different components.

Figure 49.32: Multivariate Control Chart Displaying Principal Components

For more information about star charts, see "Displaying Auxiliary Data with Stars" , or consult the entries for the STARVERTICES= and related options in Chapter 46, "Dictionary of Options."

Principal components are not the only approach that can be used to interpret multivariate control charts. This problem has recently been studied by a number of authors, including Doganaksoy and others (1991), Hawkins (1991, 1993), and Mason and others (1993).

Chapter Contents
Previous
Next
Top