Chapter 11. Data Plots and Distributions
What you will learn in this chapter of the book is to plot data and to spot opportunities from these plots. You will also learn what a "normal" distribution of data is and some terminology to describe this distribution. As in the previous section, this information will help you to solve many real problems and is needed for Six Sigma work. Plotting data is a necessary step in implementing many of the Six Sigma tools. It is used in all of the steps of the DMAIC methodology.
NOTE
Charts (Histograms) and the Normal Distribution
Manufacturing Make histograms on processes, plants, or shifts that are supposed to be the same and you may find differences.
Sales Use histogram charts to compare sales results year to year, by month, or among sales offices. You can also compare success in getting new customers or graph additional sales generated per travel dollar.
Marketing Look at market segments per month and compare changes year to year with marketing campaigns or dollars spent on advertising.
Accounting and Software Development Use histogram charts to graphically compare employees in terms of lines of code written or accounting forms completed. These groups are very visually oriented and relate well to data presented in this way.
Receivables Plot the monthly receivables year over year and compare it with cash flow.
Insurance Compare surgeries done in comparable hospitals to spot cost differences.
Case Study: Comparing Plots of Two Production LinesAproduction plant had two similar lines producing containers. The wall thickness on these containers was critical to the customer, so measurements were taken regularly and entered into a computer file. The customer had periodically expressed preference for containers from line #2 over containers from line #1, but the customer had no data to substantiate this preference. Since both lines made product within specifications, the container plant felt that any difference was imagined, since both lines were thought to be identical. Finally, an engineer plotted 1000 random wall thickness measurements from each line. Figure 11-1 shows the histogram plots of the data from the two lines, with one histogram overlaying the other for ease of comparison. Figure 11-1. Histogram of two production lines
As you can see by looking at the above histograms, the wall thickness measurements from line #1 are more dispersed (more high and low values) than those from line #2. This is why the customer was happier with the containers from line #2. Using the above data as a motivator, the engineer was able to find subtle differences between the two lines and then eliminate those differences. The wall thickness of containers from line #1 became nearly identical to the wall thickness of containers from line #2. The customer saw $25,000 per year savings from the resultant improved product and continuing plots of the data after the changes substantiated that the two lines were now making nearly identical product. |
In this case study, there are several things of note. First, the customer's feelings on quality were ignored because there was no supportive data and both lines were making product within specification. Second, the data were already available in a database that no one had bothered to examine. Third, although both lines made product within specifications, the customer saw the improvement in the revised process. Fourth, without anyone realizing it, the two lines were not identical and over time small changes had been incorporated and had not been documented. Once someone decided to plot the data, it was obvious that the customer was correct and that the products from the two lines were not the same.