JMPer Cable Blog
A technical blog for JMP users of all levels, full of how-to's, tips and tricks, and detailed information on JMP features
Measurement Systems Analysis Dashboard add-in
Measurement systems analysis (MSA)
Industrial manufacturing processes often have many sources of variability and tight tolerances. Therefore, understanding the variability behind the data is often a critical step to improving a process.
Measurement systems analysis is the broad category of statistical tools and methods designed to analyze the variability of a process. Particularly, gauge R&R is a methodology that separates variability into distinct categories. These categories are repeatability and reproducibility, and they help measure how consistent a process is through both identical and varied conditions.
The Measurement Systems Analysis Dashboard add-in for JMP 19 or later is designed to give users a convenient glimpse into their measurement system. This add-in creates a streamlined and customizable dashboard that can easily be shared in a variety of ways to be most convenient to the user. The add-in uses a targeted selection of the elements found in JMP’s broader MSA platform to provide visualizations and statistics that can efficiently and effectively represent the variability of a given process.
The add-in can be found at MSA Dashboard: JMP Marketplace.
The MSA Dashboard add-in
This section explains each element of the MSA Dashboard add-in using the Gasket.jmp (Wheeler 2006) sample data as an example. A data table with relevant scripts can be found attached to this blog.
Dialog
First, to access the initial dialog, go to the add-in menu and select MSA Dashboard to see the initial dialog. The dialog is designed to be similar to JMP’s MSA platform. To proceed, only two columns are required. At least one continuous column needs to be assigned as the Y/Measurement, and one column needs to be assigned as the Part/Sample ID variable. Optionally, one or multiple grouping columns can be selected.
Next, there are several other options that are available. First, the user can select the appropriate model type for their study design. Available options are crossed or nested. There are also options for using either standard deviation or range for the dispersion chart, as well as for selecting the alpha level and the sigma multiplier that are used in some of the calculations.
The app also has options for adding or editing optional MSA metadata (tolerance range, tolerance limits, historical mean, and historical sigma) that you may be part of your process. By default, with nothing selected in the MSA metadata section, the MSA Dashboard will attempt to use a saved MSA column property to get MSA metadata for the process, such as a historical sigma and tolerance range. However, this can be changed in two ways. First, the user can check the Launch Metadata Dialog box, which launches an extra dialog once the user has clicked OK in the initial dialog. The metadata dialog allows the user to input a tolerance range through multiple methods, including a range, a two-sided limit, or a one-sided limit. The metadata dialog is automatically populated with any metadata already in the MSA column property. Additionally, if the user checks the Use Spec Limits for Tolerance box, the tolerance limits and range use any specification limits that have been saved to the Spec Limits column property.
Lastly, there is an option to show or hide all the available control panels for the graphs within the dashboard upon launch. After all the desired options are selected and the required columns are set, clicking OK begins the process of generating the dashboard.
Overall layout
The MSA Dashboard is primarily composed of a grid of graphs that give a quick and powerful insight into the measurement system of the given process. Options that affect the entire dashboard can be found in the primary red triangle menu. Options for individual components can be found in the red triangle menu of the respective outline box containing the component.
Variance component bar chart
The variance component bar chart, found in the top left, is designed to display the sources of variability in the measurement system. It has several options in its control panel or red triangle menu. These options are grouped into three primary categories:
- Graphic Options.
- Metrics.
- Percentage Metrics.
The Graphic Options category includes toggling Labels to show or hide the values for each bar in the graph; toggling Detail Bars to expand the reproducibility bar into the individual components or reduce them back to the reproducibility bar; and toggling Total Bar to show or hide a bar that represents the total process variation.
The Metrics category includes all metric options except for percentage-based metric ones. These options include the Variance Component, Std Dev (standard deviations), and Variation (6*StdDev). None of these are turned on by default at initial launch.
The Percentage Metrics category includes all percentage-based metric options, including % Total Var Comp (variance components), % Std Dev (standard deviations), % Tolerance, and % Process. The % Total Var Comp option is selected by default at launch. The % Tolerance option is only available if a tolerance range or limits have been provided, and the % Process option is only available if a historical sigma has been provided.
All options can be used together except for Metrics and Percentage Metrics. When selecting an option in either metric category, the options in the other category will be automatically disabled.
Average chart
The average chart, found beneath the variance component bar chart, is a control chart designed to help visualize the observed variability. The average chart shows the part averages for every combination with control limits based on all the data. These limits can be used to help determine if your measurement system is good enough to detect differences between parts. 
Dispersion chart
The dispersion chart, found beneath the average chart, is a control chart that shows either the ranges or standard deviations calculated from each set of measurement replicates. It is used to help visualize whether the variability between replicates is consistent across the different variables. Both the average chart and dispersion chart do not have customization options.
Grouping chart
The grouping chart, found in the top-right corner, is designed to display the individual points by a grouping variable if one was provided. If no grouping variable is provided, it displays all of the points together. The user can switch the grouping variable that’s used if multiple grouping variables were provided via the control panel or the red triangle menu. Additionally, the graph can be altered by toggling the Graphic Options: Points, Box Plot, and Connected Means. 
Part chart
The part chart, found beneath the grouping chart, is designed to highlight the overall trend across parts. It does this by having a line that connects the mean of each part. It also colors the points by the grouping variable if it is provided. If more than one grouping variable is provided, the user can determine which grouping variable is used to color the points in the graph through the control panel or the red triangle menu.
Interaction plot
The interaction plot, found beneath the part chart, is designed to show the interaction between the grouping variables and the part variable. The interaction plot only appears for crossed designs with at least one grouping variable. The grouping variable is used as the overlay variable. Similar to the part chart, if more than one grouping variable is provided, the user may determine the grouping variable used for the overlay through the control panel or the red triangle menu. 
Summary report
The summary report is hidden by default but can be accessed through the primary red triangle menu. The summary report appears beneath the graphs when turned on. The summary report has its own red triangle menu to access different options. By default, AIAG gauge R&R results are shown, but EMP gauge R&R results and EMP results are also available.
Custom description
An editable description text box is available at the top of the MSA Dashboard report. The description text box can be accessed by selecting Edit Description in the primary red triangle menu of the dashboard. Selecting it initially adds autogenerated description text to the top of the dashboard in an editable text box. The autogenerated description contains information about the study design and can be freely edited. After completing the desired edits, the user can click the Set Description button to switch the description to a static view. If the user wishes to make further edits or hide the description, then the options to do so reappear if the user selects Edit Description from the primary red triangle menu. Selecting Hide Description hides the description from view while keeping any edits that were made. 
Save options
The primary red triangle menu also has multiple options for saving the report, including:
- Save as PDF.
- Save as picture.
- Save as interactive HTML.
- Copy to clipboard.
- Save script to data table.
- Save script to script window.
The first four options temporarily hide the graph control panels when saving. The last two options save all changes that were made through either the red triangle menus or control panels as a JSL script. However, changes that were made outside of that, such as resizing or options found through right-clicking, are not saved to the JSL script.
Additional resources
An example data table with a customized dashboard saved to the script can be found in the attached file: Gasket - MSA Dashboard Blog Example.jmp.
For more measurement systems analysis needs, try the MSA platform in JMP. The full MSA platform includes additional tools, analyses, and options for evaluating measurement systems, including support for more complex models.
To learn about measurement systems analysis, try JMP’s free Statistical Thinking for Industrial Problem Solving (STIPS) course or visit JMP’s Statistics Knowledge Portal to learn more about MSA and other statistical methods. MSA can be found under Chapter 3: Quality Methods in STIPS.
References
Donald J. Wheeler (2006). EMP III Evaluating the Measurement Process & Using Imperfect Data. Second Printing, SPC Press, Inc., Knoxville, TN
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.