[See SHWXS3 in the SAS/QC Sample Library]You can use output data sets from a number of SAS procedures as input data sets for the SHEWHART procedure. In this example, the MEANS procedure is used to create a data set containing subgroup summary statistics, which can be read by the SHEWHART procedure as a HISTORY= data set. The following statements create an output data set named Oilsummeans, which contains subgroup means, standard deviations, and sample sizes for the variable KWatts in the data set Turbine (see Creating Charts for Means and Standard Deviations from Raw Data):
proc means data=Turbine noprint; var KWatts; by Day; output out=Oilsummeans mean=means std=stds n=sizes; run;
A listing of Oilsummeans is shown in Output 15.42.1.
Summary Statistics for Power Output Data |
Day | _TYPE_ | _FREQ_ | means | stds | sizes |
---|---|---|---|---|---|
04JUL | 0 | 20 | 3487.40 | 220.260 | 20 |
05JUL | 0 | 20 | 3471.65 | 210.427 | 20 |
06JUL | 0 | 20 | 3488.30 | 147.025 | 20 |
07JUL | 0 | 20 | 3434.20 | 157.637 | 20 |
08JUL | 0 | 20 | 3475.80 | 258.949 | 20 |
09JUL | 0 | 20 | 3518.10 | 211.566 | 20 |
10JUL | 0 | 20 | 3492.65 | 193.779 | 20 |
11JUL | 0 | 20 | 3496.40 | 212.024 | 20 |
12JUL | 0 | 20 | 3398.50 | 199.201 | 20 |
13JUL | 0 | 20 | 3456.05 | 173.455 | 20 |
14JUL | 0 | 20 | 3493.60 | 187.465 | 20 |
15JUL | 0 | 20 | 3563.30 | 205.472 | 20 |
16JUL | 0 | 20 | 3519.05 | 173.676 | 20 |
17JUL | 0 | 20 | 3474.20 | 200.576 | 20 |
18JUL | 0 | 20 | 3443.60 | 222.084 | 20 |
19JUL | 0 | 20 | 3586.35 | 185.724 | 20 |
20JUL | 0 | 20 | 3486.45 | 223.474 | 20 |
21JUL | 0 | 20 | 3492.90 | 145.267 | 20 |
22JUL | 0 | 20 | 3432.80 | 190.994 | 20 |
23JUL | 0 | 20 | 3496.90 | 208.858 | 20 |
The variables MEANS, STDS, and SIZES do not follow the naming convention required for HISTORY= data sets (see HISTORY= Data Set. The following statements temporarily rename these variables to KWattsX, KWattsS, and KWattsN, respectively (the names required when the process KWatts is specified in the XSCHART statement):
title 'Mean and Standard Deviation Charts for Power Output'; proc shewhart history=Oilsummeans (rename=(means = KWattsX stds = KWattsS sizes = KWattsN )); xschart KWatts*Day; run;
The resulting charts are identical to the charts in Figure 15.116.