Previous Page | Next Page

The CUSUM Procedure

Reading Cusum Scheme Parameters

[See CUSONES1 in the SAS/QC Sample Library]This example shows how the cusum parameters saved in the previous example can be applied to new measurements saved in a data set named Cans2:

data Cans2;
   length pump $ 8;
   label Hour = 'Hour';
   input Hour Weight pump $ 8. ;
datalines;
16  8.1765  Pump 3
17  8.0949  Pump 3
18  8.1393  Pump 3
19  8.1491  Pump 3
20  8.0473  Pump 1
21  8.1602  Pump 1
22  8.0633  Pump 1
23  8.0921  Pump 1
24  8.1573  Pump 1
25  8.1304  Pump 1
26  8.0979  Pump 1
27  8.2407  Pump 1
28  8.0730  Pump 1
29  8.0986  Pump 2
30  8.0785  Pump 2
31  8.2308  Pump 2
32  8.0986  Pump 2
33  8.0782  Pump 2
34  8.1435  Pump 2
35  8.0666  Pump 2
run;

The following statements create a one-sided cusum chart for the measurements in Cans2 using the parameters in cusparm:

ods graphics on;
symbol h = 1;
title "One-Sided Cusum Analysis for New Data";
proc cusum data=Cans2 limits=cusparm;
   xchart Weight*Hour( pump ) / odstitle=title;
   label Weight = 'Cusum of Weight';
run;
ods graphics off;

ODS Graphics remains enabled until it is disabled with the ODS GRAPHICS OFF statement, so this cusum chart is also created using ODS Graphics.

The LIMITS= option in the PROC CUSUM statement specifies the data set containing preestablished cusum parameters.1 The chart, shown in Figure 6.12, indicates that the process is in control. Levels of the variable pump (referred to as a block-variable) do not enter into the analysis but are displayed in a block legend across the top of the chart. See Block Variable Legend Options.

Figure 6.12 Cusum Chart with Decision Interval for New Data
Cusum Chart with Decision Interval for New Data

In general, the parameters for a specified process and subgroup-variable are read from the first observation in the LIMITS= data set for which

  • the value of _VAR_ matches the process (in this case, Weight)

  • the value of _SUBGRP_ matches the subgroup-variable name (in this case, Hour)

If you are maintaining more than one set of cusum parameters for a particular process, you will find it convenient to include a special identifier variable named _INDEX_ in the LIMITS= data set. This must be a character variable of length 16. Then, if you specify READINDEX=’value’ in the XCHART statement, the parameters for a specified process and subgroup-variable are read from the first observation in the LIMITS= data set for which

  • the value of _VAR_ matches process

  • the value of _SUBGRP_ matches the subgroup-variable name

  • the value of _INDEX_ matches value

In this example, the LIMITS= data set was created in a previous run of the CUSUM procedure. You can also create a LIMITS= data set with the DATA step. See LIMITS= Data Set for details concerning the variables that you must provide.

Previous Page | Next Page | Top of Page