The CUSUM Procedure |

[See CUSTWOS1 in the SAS/QC Sample Library]The previous example illustrates how you can create a cusum chart using raw process measurements read from a DATA= data set. In many applications, however, the data are provided in *summarized form* as subgroup means. This example illustrates the use of the XCHART statement when the input data set is a HISTORY= data set.

The following data set provides the subgroup means, standard deviations, and sample sizes corresponding to the variable Weight in the data set Oil (see the section Creating a V-Mask Cusum Chart from Raw Data:

data Oilstat; label Hour = 'Hour'; input Hour WeightX WeightS WeightN; datalines; 1 8.0938 0.0596 4 2 8.0925 0.0902 4 3 8.1010 0.0763 4 4 8.1198 0.0256 4 5 8.1013 0.0265 4 6 8.0800 0.0756 4 7 8.1145 0.0372 4 8 8.0830 0.0593 4 9 8.0618 0.0057 4 10 8.1023 0.0465 4 11 8.1065 0.0405 4 12 8.0993 0.0561 4 ; run;

The data set Oilstat is listed in Figure 6.4.

Since the data set contains a subgroup variable, a mean variable, a standard deviation variable, and a sample size variable, it can be read as a HISTORY= data set. Note that the names WeightX, WeightS, and WeightN satisfy the naming conventions for summary variables since they begin with a common prefix (Weight) and end with the suffix letters *X*, *S*, and *N*.

The following statements create the cusum chart:

title 'Cusum Chart for Average Weights of Cans'; proc cusum history=Oilstat; xchart Weight*Hour / mu0 = 8.100 /* target mean */ sigma0 = 0.050 /* known standard deviation */ delta = 1 /* shift to be detected */ alpha = 0.10 /* Type 1 error probability */ vaxis = -5 to 3 ; label WeightX = 'Cumulative Sum'; run;

Note that the *process* Weight specified in the XCHART statement is the prefix of the summary variable names in Oilstat. Also note that the vertical axis label is specified by associating a variable label with the subgroup mean variable (WeightX). The chart (not shown here) is identical to the one in Figure 6.2.

In general, a HISTORY= input data set used with the XRCHART statement must contain the following four variables:

subgroup variable

subgroup mean variable

subgroup range variable

subgroup sample size variable

Furthermore, the names of subgroup mean, standard deviation, and sample size variables must begin with the prefix *process* specified in the XRCHART statement and end with the special suffix characters *X*, *S*, and *N*, respectively.

Note that the interpretation of *process* depends on the input data set specified in the PROC CUSUM statement.

If raw data are read using the DATA= option (as in the previous example),

*process*is the name of the SAS variable containing the process measurements.If summary data are read using the HISTORY= option (as in this example),

*process*is the common prefix for the names containing the summary statistics.

For more information, see DATA= Data Set and HISTORY= Data Set.

Copyright © SAS Institute, Inc. All Rights Reserved.