Our World Statistics Day conversations have been a great reminder of how much statistics can inform our lives. Do you have an example of how statistics has made a difference in your life? Share your story with the Community!
JMP user Archana Pawse, PhD, is a Six Sigma Black Belt who has worked for Applied Magnetics, Superconductor Technologies and JDS Uniphase as a Product/Process/Reliability Engineer.
What follows is Dr. Pawse's description of how she uses JMP in her line of work, which she wanted to share with others. Feel free to leave a question or comment for her to respond to or contact her directly by e-mail.
How I Use JMP for Root Cause Analysis
In a manufacturing environment, it is very critical to find the root cause of failures quickly. Delay in identifying the root cause can result in wasted money and resources. JMP is useful for root cause investigation because you can easily explore and graph the data in multiple ways. I have found three JMP features/techniques -- selecting cells, partition plots and variability charts -- to be useful in this investigation.
To demonstrate these techniques, I have generated some hypothetical data. In this example, Y1, Y2 and Y3 are performance parameters for a product that is made using three processes P1, P2 and P3.
P1-Machine, P1-X1 and P1-X2 are the parameters for process P1. P2-Machine, P2-X1, P2-X2 are the parameters for process P2. P3-Machine, P3-X1 are the parameters for process P3. Parameter Y3 has 15 failures (value <= 0.7).
1. Selecting cells
In JMP, once you have graphed all relevant parameters, you can select failed data points on a graph or cells in the table, and they are highlighted on all the graphs in the session. This option is not available in some other software. By highlighting the failures, you can immediately see if there is any correlation between various parameters or if outliers on one graph are also outliers on other graphs.
In this example, I have graphed distributions of parameters Y1, Y2 and Y3, and I have selected all the failures for the parameter Y3. These failures are also highlighted on distributions of Y1 and Y2, which show that the failures are also outliers for parameter Y1. This kind of information can give more insight into the cause of failures.
2. Partition plots
If your final product performance depends on various process parameters and its interactions, then it is very time-consuming to review all process parameters to find the cause of a product failure. Partition analysis can narrow this list down to a few parameters, so you can investigate these few parameters in detail. In some cases, the partition analysis can even identify the root cause.
In this example, the Y3 parameter is the response variable, and all the process parameters are entered in the X factor field. The best split option in partition analysis shows that the product fails (has a low Y3 value) when it is processed in machine B at P2 process and machine 3 at the P1 process.
3. Variability charts
Variability charts are useful for visualizing failures. This graph is very effective at displaying interactions between various parameters because you can plot Y versus multiple X parameters. In addition, by using multiple colors, symbols, symbol sizes, you can show at least four parameters on this graph.
In this example, it’s easy to see from the graph that all failed product (Y3<=0.7) is made using machine B at step P1 and machine 3 at step P2. In addition, by color-coding the data for the different P3 machines, you can see that these failures are not related to the machine used at step P3. Different symbols are used for the passed and failed units for easier identification.
The techniques/features described here are not limited only to root cause investigation. They can also be used for other manufacturing applications like process optimizations as well as in marketing applications to identify a target market or cross-sell opportunities.