Using Pareto charts in project quality control

Using Pareto charts in project quality control

When you're analyzing problems on a project, or measuring the impact of procedural changes, you need a tool that will pinpoint the important areas quickly. Pareto charts are easy to generate once you have chosen the data to be analyzed. Even those who are unfamiliar with data analysis can easily understand them - so it can be useful to include them in reports and presentations.

What is a Pareto chart?

A well-designed Pareto chart should provide you with an instant analysis of a situation

A Pareto chart is a bar chart that's organized according to specific guidelines. The bars are arranged in descending order from left to right. You use it to quickly identify the sources of problems and to see how individual pieces of data fit into a whole process. When you implement a solution to a problem, you can use a series of Pareto charts as a comparative tool to assess the effect of the solution. A well-designed Pareto chart should provide you with an instant analysis of a situation.

The "80 percent/20 percent" rule

The Pareto chart is based around the "80 percent/20 percent" rule. This is a phenomenon that was first noted by an Italian economist named Vilfredo Pareto (1848 - 1923), who observed that 80 percent of a country's wealth is usually held by 20 percent of the population. This principle, known as the Pareto effect, was applied to other situations and a similar pattern was observed. The proportions may vary, but the principle of a small number of events or conditions (the "vital few") influencing a greater number of outcomes (the "useful many") holds true in many areas, including quality management. It's a useful principle for determining the cause of problems, and a good way of checking the relevance of the data that you include in your Pareto chart. If the chart doesn't produce a clear result, you might need to regenerate it using different data.

Choosing the right data

Before you draw the chart, you need to decide on the data to be analyzed. Let's say you're analyzing data in a production process where six machines must produce a weekly quota of 3,000 units. The acceptable failure rate is 30 units per week, but when you study your first report you discover an average failure rate of 120 units per week.

You begin by grouping the failures according to the following major faults:

• faulty soldering
• missing jumpers
• bent pins
• incorrect switch settings

And you construct your first Pareto chart based on these results.

A quick look at the Analysis by fault type chart tells you that no one fault type is responsible for the large number of failures in the process. If you take steps to eliminate all of the soldering faults, for example, the process still produces eighty faulty units per week, more than twice the acceptable failure rate.

You decide to redesign the chart using different data. This time, you analyze the number of faulty units produced by a particular machine to see if that yields a more meaningful result.

You can see from the Analysis by machine number chart that the vast majority of faulty units are produced by machine number 4. You can analyze this data further to determine whether the high number of faults is due to a problem with the machine, or perhaps an inexperienced crew. You can use the charts to illustrate your findings in reports or presentations on the matter.

Drawing the chart

As you can see, the Pareto chart is laid out with the vital few, or 20 percent, on the far left of the chart, and the numbers decrease as you read across the chart from left to right, with the lowest number to the far right of the chart. Sometimes you might find it useful to include a cumulative curve to reinforce a pattern. Cumulative curves are usually found on comparative Pareto charts. In the example shown above, when the problem is diagnosed and fixed, you can use a series of before-and-after charts to display the improvements in a graph.