Share your ideas for the JMP Scripting Unsession at Discovery Summit by September 17th. We hope to see you there!
Choose Language Hide Translation Bar
Community Manager Community Manager

Identifying Poorly Performing Processes

Learn more in our free online course:
Statistical Thinking for Industrial Problem Solving


In this video, we show how to identify poorly performing processes in JMP using the Semiconductor Capability data, found in the Sample Data library.


This table has 128 process variables grouped together in the columns panel. Each variable has the spec limits saved as a column property.


First, we conduct a capability analysis for the first 10 variables.


To conduct the analysis, we select Process Capability from the Analyze menu under Quality and Process.


We select the first 10 variables, from NPN1 to NPN4, and click Y, Process.


Notice that there are options to define whether data were collected using rational subgrouping, to specify moving range options, to enter historical information, and to specify the distribution.


We select OK to conduct the analysis.


The triangle on the goal plot is for variables with a Ppk of 1.0 or greater.


We change the Ppk to 1.3 instead of 1.0. This adjusts the triangle to show variables with Ppk greater than 1.3 inside the triangle.


We can see that four variables fall well outside this region.


To color-code the goal plot, we select Shade Levels from the red triangle next to Goal Plot. The best performing variables are within the green triangle. Borderline processes are in the yellow triangle, and poorly performing processes are in the red zone.


When we place the mouse pointer on a variable, JMP displays the variable name. It also shows a control chart graphlet for the variable. The two most problematic variables are IVP2 (on the top left) and IVP1 (on the far right).


We scroll down to see capability box plots for each variable. Each variable is centered by its target and scaled by its spec limit. This enables us to easily identify variables that are off target or have too much variability.


We see that IVP1 is too variable and is off target. It's shifted toward the upper spec limit. IVP2 is also too variable and is shifted toward the lower spec limit. This confirms what we see on the goal plot.


The capability index plot shows the Ppk for each variable. This is another graph that helps us easily identify poorly performing variables.


Additional summaries and graphical displays are available under the top red triangle.


We select the Individual Details report. This provides individual capability analyses for each variable. We can clearly see that IVP1 is shifted well outside the spec limits and that the process spread is much wider than the width of the spec limits.


You can also generate a goal plot, and other performance statistics for many variables, using the Process Screening platform.


This platform is on the Analyze menu, under Screening.

Again, we select the first 10 variables.


The stability index, Ppk, and Cpk are provided, along with many other measures.


Process Performance Graph and Goal Plot are red triangle options.


To identify off-target processes, we select Target Index from Show Capability under the red triangle menu. We also select Cp from this menu.


Four of the ten processes are off target by 1 standard deviation or more.


We click the Target Index label twice to sort in descending order.


The variable NPN3 is stable, and it is highly capable but off target. How is this possible? We select the variable and then select Process Capability for Selected Item. The process spread is very narrow relative to the spec limits. So, despite the process being off target, the predicted nonconformance rate is 0.00.

Article Tags