Chapter Contents
Chapter Contents
Previous
Previous
Next
Next
HBAR Statement

Creating a Pareto Chart from Raw Data

In the fabrication of integrated circuits, common causes of failures include improper doping, corrosion, surface contamination, silicon defects, metallization, and oxide defects. The causes of 31 failures were recorded in a SAS data set called FAILURE1.
   data failure1;
      length cause $ 16;
      label  cause = 'Cause of Failure';
      input  cause;
      datalines;
   Corrosion
   Oxide Defect
   Contamination
   Oxide Defect
   Oxide Defect
   Miscellaneous
   Oxide Defect
   Contamination
   Metallization
   Oxide Defect
   Contamination
   Contamination
   Oxide Defect
   Contamination
   Contamination
   Contamination
   Corrosion
   Silicon Defect
   Miscellaneous
   Contamination
   Contamination
   Contamination
   Miscellaneous
   Contamination
   Contamination
   Doping
   Oxide Defect
   Oxide Defect
   Metallization
   Contamination
   Contamination
   ;

Each of the 31 observations corresponds to a different circuit, and the value of CAUSE provides the cause for the failure. These are raw data in the sense that there is more than one observation with the same value of CAUSE, and the observations are not sorted by CAUSE. The following statements produce a basic Pareto chart for the failures:

   proc pareto data=failure1;
      hbar cause / cframe   = ligr
                   cbars    = vigb
                   cconnect = salmon;
   run;

The PARETO procedure is invoked with the first statement, referred to as the PROC statement. You specify the process variable to be analyzed in the HBAR statement. The Pareto chart is shown in Figure 27.1.

parhgs1.gif (3752 bytes)

Figure 27.1: Pareto Chart for IC Failures in the Data Set FAILURE1

The procedure has classified the values of CAUSE into seven distinct categories (levels). The bars represent the percent of failures in each category, and they are arranged in decreasing order. Thus, the most frequently occurring category is Contamination, which accounts for 45% of the failures. The Pareto curve indicates the cumulative percent of failures from top to bottom; for example, Contamination and Oxide together account for 71% of the failures. If there is sufficient space, the procedure labels the bars along the vertical axis as in Figure 27.1. Otherwise the procedure numbers the bars from top to bottom and adds a legend identifying the categories.

A category legend is likely to be introduced when

Chapter Contents
Chapter Contents
Previous
Previous
Next
Next
Top
Top

Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.