A histogram is, as its name implies, a look at the history of a process. It cannot be used to predict the future as a control chart enables us to do, but it is often useful.
Let's use as an example the volume of a retail store. Data has been collected on gross sales by day for the previous year and totals $1,000,000. To keep it simple, we will arbitrarily select 10 class sizes in which to group our data.
The first bar or class size will be 0-$100,000, the second $100,001 to $200,000, etc. Then from the data, determine how many days sales were in class one, how many in class two, etc. and that determines the height of the bar in each class.
Dots in the center of the tops of the bars can be connected to show the picture of the variation in great days versus poor days. In a normal process, this line from the connection of the dots will resemble the outline or shape of a bell and is called a bell curve and indicates normal variation.
If it is skewed in any way, it presents information showing something is happening to the process that is not normal and should be investigated. The histogram is widely used in manufacturing to show variation in a process.
ADVERTISEMENT
Another variation of the bar chart is the Pareto diagram. It is named after the Italian economist, Wilfredo Paredo, who in 1897 presented a formula showing that the distribution of income is uneven. A similar theory was expressed by the U.S. economist, M.C. Lorenz, in 1907.
Both of these scholars pointed out that by far the largest share of income or wealth is held by a very small number of people. We sometimes refer to this as the 80-20 rule. Later, J. M. Juran applied this formula in order to classify problems of quality into the vital few and the trivial many, and named the method Pareto Analysis.
There are many aspects of business that could be improved: defects, time allocation, cost savings, etc. that it is difficult to know just where to begin solving the problems. Looking at these problems in the form of a loss, it is important to clarify the distribution pattern of the loss. Most of the loss will be due to a very few types of causes. By leaving aside for the time being the trivial many and focusing on solving the vital few we can be much more effective in the investment of our time and efforts.
The first step in creating a Pareto chart is to decide what problem to investigate and how to collect the data such as defects, losses in monetary terms, accidents, etc. and design a data tally sheet, gather the data, and calculate the totals by classifications of the data.
For example if you are measuring monetary losses, you might have misplaced inventory, damaged inventory, theft, obsolescence, unsalable merchandise, and other. You should always show the other in the last column even if it is greater that the other categories.
Plot the data in bar chart form with the highest loss category to the left, second highest next, and so forth until the Other category is last. This points out graphically the most costly category and where you should start attacking the problem.
It is surprising how many times 80 percent of our problems are caused by 20 per cent of the causes.
Louis Schultz, managing director of Process Management LLC, has assisted organizations worldwide with performance improvement. He currently assists area business owners as a SCORE counselor. E-mail him with questions or comments at lou@processman agement.com.
