Creating np Charts from Summary Data

[See SHWNP1 in the SAS/QC Sample Library]The previous example illustrates how you can create charts using raw data (counts of nonconforming items). However, in many applications, the data are provided in summarized form as proportions or percentages of nonconforming items. This example illustrates how you can use the NPCHART statement with data of this type.

The following data set provides the data from the preceding example in summarized form:

data Cirprop;
   input Batch pFailed @@;
   sizes=500;
datalines;
 1  0.010   2  0.012   3  0.022   4  0.012   5  0.008
 6  0.018   7  0.034   8  0.020   9  0.024  10  0.018
11  0.016  12  0.014  13  0.014  14  0.030  15  0.016
16  0.036  17  0.024  18  0.032  19  0.008  20  0.014
21  0.034  22  0.024  23  0.016  24  0.014  25  0.030
26  0.012  27  0.016  28  0.024  29  0.014  30  0.018
;

A partial listing of Cirprop is shown in Figure 15.54. The subgroups are still indexed by Batch. The variable pFailed contains the proportions of nonconforming items, and the variable Sampsize contains the subgroup sample sizes.

Figure 15.54 The Data Set Cirprop
Subgroup Proportions of Nonconforming Items

Batch pFailed sizes
1 0.010 500
2 0.012 500
3 0.022 500
4 0.012 500
5 0.008 500

The following statements create an chart identical to the one in Figure 15.53:

title 'np Chart for the Number of Failing Circuits';
symbol v=dot;
proc shewhart data=Cirprop;
   npchart pFailed*Batch / subgroupn=Sampsize
                           dataunit =proportion;
label pFailed = 'Number of FAIL';
run;

The DATAUNIT= option specifies that the values of the process (pFailed) are proportions of nonconforming items. By default, the values of the process are assumed to be counts of nonconforming items (see the previous example).

Alternatively, you can read the data set Cirprop by specifying it as a HISTORY= data set in the PROC SHEWHART statement. A HISTORY= data set used with the NPCHART statement must contain the following variables:

  • subgroup variable

  • subgroup proportion of nonconforming items variable

  • subgroup sample size variable

Furthermore, the names of the subgroup proportion and sample size variables must begin with the process name specified in the NPCHART statement and end with the special suffix characters and , respectively.

To specify Cirprop as a HISTORY= data set and Fail as the process, you must rename the variables pFailed and Sampsize to FailP and FailN, respectively. The following statements temporarily rename pFailed and Sampsize for the duration of the procedure step:

options nogstyle;
goptions ftext=swiss;
title 'np Chart for the Number of Failing Circuits';
proc shewhart history=Cirprop(rename=(pFailed =FailP
                              sizes=FailN ));
   npchart Fail*Batch / cframe    = vibg
                        cinfill   = vlibg
                        coutfill  = salmon
                        cconnect  = salmon;
run;
options gstyle;

The NOGSTYLE system option causes ODS styles not to affect traditional graphics. Instead, the NPCHART statement options control the appearance of the graph. The GSTYLE system option restores the use of ODS styles for traditional graphics produced subsequently. The resulting chart is shown in Figure 15.55.

Figure 15.55 Chart for Circuit Failures (Traditional Graphics with NOGSTYLE)
np Chart for Circuit Failures (Traditional Graphics with NOGSTYLE)

In this example, it is more convenient to use Cirprop as a DATA= data set than as a HISTORY= data set. As illustrated in the next example, it is generally more convenient to use the HISTORY= option for input data sets that have been created previously by the SHEWHART procedure as OUTHISTORY= data sets.

For more information, see HISTORY= Data Set.