The Power of Pareto
The Pareto chart (not to be directly confused with the Pareto probability distribution function) is a simple approach to revealing significance in data. Before we plot our chart, we need to complete some initial work:
- Gather the data in a natural format (count, floating point [decimal], dollars, etc.)
- Sort the data from high to low using the meaningful natural format
- Calculate the sum of the values in the chosen format
- Convert the individual values to percentage by dividing by the sum of the values and multiplying by 100 (or convert to “%” using a spreadsheet)
- Plot the ordinal high to low values on the left y-axis using a histogram
- Plot the percentage on a second y-axis (right side) using a line diagram
- Congratulations, you now have a Pareto chart
The power of the Pareto chart lies in how it makes the significance of the first few items completely obvious. The so-called 80/20 rule is derived from this approach, since the early 20th century Italian economist, Vilfredo Pareto, supposedly observe that 20% of the people have 80% of the wealth in any given country. An alternative approach is Zipf’s Law, which suggests unusual events occur rarely while common events occur frequently (hardly a revelation!). The point of Zipf’s Law is that is allows us to express our empirical data with a mathematical power law that strongly resembles the insight of Pareto.
With situation like warranty, we may want to plot both the number of returns of a given product as well as the dollar cost of the return. The dollar value will give us a better idea of pain levels. Communication with the customer will help us set priorities.
Always be sure the understand any potential distortions; for example, if we are charting a measurement that doesn’t really mean anything, we learn nothing. If we are charting a surrogate (substitute) value because that is all we can measure, let us verify that it means what we think it means.