10.07 Pareto charts

A Pareto chart is used to identify the most significant factors in a set of categorical data. The chart combines a column graph and a line graph, and has two vertical axes, one for each graph type.

• The columns represent the frequency for each category, or factor in a process, but could also represent cost, or some other unit of measurement. The columns are arranged in descending order, from tallest on the left to shortest on the right (i.e. most significant to least significant). The value for each column is read from the left vertical axis.
   
• The line graph or polygon represents the cumulative percentage of the values for each column. It always curves upwards, but decreases in steepness, from left to right. The value of any point on the line graph is read from the right vertical axis.

Below is a Pareto chart showing some of the common reasons for failing a driving test. 

 

We can see from the chart above that the most common reasons for failing a driving test are:

• Observations at intersections (not checking for traffic carefully in all directions)
• Use of mirrors (not using the cars mirrors to check the position of other vehicles)

If we draw a line from the $80%$80% mark on the right vertical axis to the line graph, and then continue that line down to the horizontal axis, the most important factors appear on the left of the line. In this case, it is the first two factors (represented by the first two points on line graph) that contribute to the majority of driving test failures. One could argue that the third factor, 'inappropriate speed' is also significant, due its closeness to the $80%$80% mark. 

Pareto charts are based on something called the Pareto principle, which says that around $80%$80% of problems in a process tend to come from only $20%$20% of factors. While these percentages are only a guide, they are common enough to be called the $80$80/$20$20 rule.

Process improvement teams often use Pareto charts to determine which factors in a process are causing the most problems, so they can focus their efforts on those. This is an important part of quality control, often used to improve customer service or reduce the number of defects in a product.

Worked example

A hotel collected data on customer complaints over the course of a month and organised the data into a frequency table:

The hotel wants to display this data in a Pareto chart and present it to staff so that the most significant complaints can be addressed.

The first step is to add two additional columns to the frequency table: one for cumulative frequency and the other for cumulative percentage.

The cumulative percentage is found by dividing each cumulative frequency by the total number of complaints, then multiplying by $100$100. For example, in the first row, the cumulative percentage is $\frac{33}{80}\times100=41.3%$3380​×100=41.3% to $1$1 decimal place.

• The number of complaints (frequency) column is used to draw the vertical bars, with its corresponding axis on the left.
• The cumulative percentage column is used to draw the line graph, with its corresponding axis on the right. 

Practice questions

Question 1

Question 2

Question 3