Note: See c Chart for Varying Number of Inspection Units in the SAS/QC Sample Library.
In applications where the number of inspection units per subgroup is not equal to one, a u chart is typically used to analyze the number of nonconformities per unit (see UCHART Statement: SHEWHART Procedure). However, as shown in this example, you can use the CCHART statement to create a c chart for this type of data.
Output 17.10.1 illustrates a situation in which varying numbers of trucks are painted each day. Trucks painted on the same day are regarded
as subgroups, and each truck is regarded as an inspection unit. The following statements create a SAS data set named Trucks4
, which contains paint defects for trucks painted on 26 days:
data Trucks4; input Day Defects Ntrucks @@; label Day='Day' Defects='Number of Paint Defects'; datalines; 1 5 1 2 9 3 3 5 2 4 9 2 5 24 4 6 10 2 7 15 3 8 17 3 9 16 3 10 13 2 11 28 4 12 18 5 13 8 2 14 7 2 15 5 1 16 17 3 17 2 1 18 17 3 19 15 4 20 19 5 21 6 3 22 23 5 23 27 4 24 6 2 25 12 2 26 12 3 ;
The variable Defects
provides the defect count () for the ith day, and the variable Ntrucks
provides the number of inspection units (). The following statements create a c chart for this data:
ods graphics off; title 'c Chart for Varying Number of Units Per Subgroup'; proc shewhart data=Trucks4; cchart Defects*Day / subgroupn = Ntrucks nolegend; run;
The SUBGROUPN= option specifies the subgroup sample size variable Ntrucks
(in general, the values of this variable need not be integers). Alternatively, you can specify a fixed value with the SUBGROUPN=
option. When this option is not specified, it is assumed that .
The chart is shown in Output 17.10.2. Note that the central line and the control limits vary with the number of inspection units.