Adding Fitted Normal Curves to a Comparative Histogram

See CAPCMH1 in the SAS/QC Sample LibraryIn Figure 5.6, it appears that each lot produces transistors with channel lengths that are normally distributed. The following statements use the NORMAL option to fit a normal distribution to the data for each lot (the observations corresponding to a specific level of the classification variable are referred to as a cell). The normal parameters $\mu $ and $\sigma $ are estimated from the data for each lot, and the curves are superimposed on each component histogram.

title "Comparative Analysis of Lot Source";
proc capability data=Channel noprint;
   specs lsl = 0.8 usl = 2.0;
   comphist Length / class     = Lot
                     nrows     = 3
                     intertile = 1
   label Lot = 'Transistor Source';

The comparative histogram is displayed in Figure 5.7.

Figure 5.7: Fitting Normal Curves

Fitting Normal Curves

Specifying INTERTILE=1 inserts a space of one percent screen unit between the framed areas, which are referred to as tiles. The shaded bars, added with the CPROP= option, represent the relative frequency of observations in each cell. See Dictionary of Options for details concerning these options.