Let us assign times, investments and probabilities to each of the defects,and
then calculate the P. I. The results are shown in Table 3.
TABLE 3 (1) (2) (3) (4) Annual Time to Invest- Probability Defect Value($) Recover ment($) of Success P.I. Years C 300,000 5.8 150,000 .65 26 E 250,000 2.0 50,000 .80 200 G 130,000 3.5 80,000 .60 28 B 45,000 1.0 10,000 .85 383 D 45,000 1.5 15,000 .50 100 A 35,000 0.8 5,000 .70 613 H 20,000 2.5 8,000 .95 95 F 20,000 1.5 10,000 .70 93 J 10,000 2.0 4,000 .40 50 I 10,000 1.0 6,000 .80 133 (1) x (4) P.I. = --------- x 100 (2) x (3) It now appears that defect A, B and E should receive priority. What happened to defect C, our plum we were so eager to pluck? During the five years that the project team is working on defect C, losses have accumulated to $1.5 million. With a probability of success of only 0.65, the potential recoverable losses are $195,000 a year, but this would not begin to be realized until the end of five years. Since it cost us $150,000 to solve the problem, substantial savings won't show up until the seventh year. Is it worth waiting that long? Probably not, when you consider the rapid changes in policy, personnel and technology that take place today. A better solution might be to invest enough money in problem solving to reduce the time it would take to increase the probability of success. You might even experiment with the Priority Index formula and come up with a different combination of parameters and weighting scheme; I would encourage you to try it. We have taken a new look at our old friend, the Pareto principle, and have examined its application to cost reduction and problem solving from a different perspective. We have added new dimensions of time, investment dollars and probability of success because these parameters are compatable with modern management's approach to decision making. A Priority Index for combining these dimensions seems to have promise as an effective, realistic management tool for today and tomorrow. THE PARETO TECHNIQUE As can be seen from the preceding discussion, the Pareto analysis has application for identifying which of several problem areas deserve the greatest attention. It provides hard data on which problems occur with greatest frequency rather than the usual reliance on personal opinion based on undocumented observation. The technique is relatively simple to apply. We have pointed out that the normal curve has its greatest frequency of occurrence of an event near the center of the distribution. This is true because a frequency distribution is a classification of a group of items by some quantitative characteristic. he Pareto distribution, on the other hand, is arranged in the order of the measurement value of the elements. In other words, for the Pareto distribution we rank a series of events or problems in the order of the numeric value assigned to those events. We might use the tabulated occurrence of the problem or the dollar value associated with the problem or any other quantitative measure which might represent the reason for wanting to identify and solve the problem. Data can be presented in tabular form as was done in the preceding examples presented in the article by Hy Pitt. Another method of presentation which provides a visual means of interpretation might be the histogram or more commonly called the bar graph. In order to utilize the Pareto approach to problem solving we must first identify a series of problem areas. Each problem must be one which we can quantify in the same units (dollar losses for example). We will discuss a method of identifying these problem areas in the next section. Let's look at an example of the method of isolating the serious problem utilizing the Pareto analysis. This example is adapted from the publication, Q C Circles: Application, Tools and Theory. The basic phases of the Pareto Technique are as follows: 1. List all the problem areas associated with the situation. Include all elements at this stage so as not to rule out any area at too early a stage. 2. Measure the elements using the same measure for each. Usually this will be in terms of dollars, time, frequency, etc. At this point no order is given to the elements or to the measurements. 3. Arrange the elements according to their measurement, chart the results and graph the elements and their measurements. From the chart and graph it is possible to: a) assign priorities to the various elements b) gain an understanding of the elements c) provide a better description of the elements d) generate standard information for any analysis which follows e) select a goal using the Pareto chart as supporting documentation. If the Pareto distribution is ordered by dollar value it becomes possibleto identify areas which will yield the greatest cost benefit in attacking theproblems.4. Construct a cumulative distribution for the prioritized items andmeasurements. These cumulative distributions when charted and graphed to forma Pareto curve will permit us to see that a few items account for adisproportionate amount of all measurements.5. Study and interpret the results. A good place to begin improvementis to work on those elements whose measurements are the greatest. The decisionmust be made as to what measurements to use in the analysis as was shown in thePitt article. PARETO EXERCISE The following scrap data was collected by part number and operation. PART OPERATION NUMBER NUMBER A 40 D 70 C 50 C 70 D 20 B 70 E 20 D 50 C 70 A 40 D 20 B 70 E 50 B 70 C 20 E 70 D 40 D 70 E 70 B 70 1. Construct a Pareto chart by part number. 2. Construct a Pareto distribution by part number. 3. Construct a Pareto chart by operation number. 4. Construct a Pareto distribution by operation number. 5. What conclusions may be drawn from this analysis?