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

## Reliability Plotting ("No-Calc" Sampling Plan) - Powerful Univariate Method of R/C Estimation

Has anyone working on disposable medical products (or other single use products) ever wished that JMP had a user-friendly graphically driven "Reliability Plotting" platform or similar graphically driven method for calculating Reliability and Confidence (R&C) directly for non-normal data?

Reliability Plotting is a method originally developed with industrial application in electronics, and is a way of applying transformations to univariate data on a plot of Median Rank (or % Cumulative of the Data, transformed if necessary) versus the Raw data (or a transformation of that data, if necessary) -- in order to linearize that plot for the purposes of R&C estimation (where R in this context refers to % of product in-specification and C refers to % confidence level).

So, ultimately transformations are applied in Y and/or in X to get a best-fitting straight line, and the % in-spec can be determined at the extrapolated value of Y at the corresponding specification limit(s) in X.  From this extrapolation a confidence limit can be calculated, and the resultant %in-spec (i.e. % Reliability) at the chosen % Confidence level can be obtained.

This method is very powerful because it uses more information than k-tables for computing C&R, where it considers the relative location of the data to the line of best fit, whereas the classical "k-tables method" uses only the mean and standard deviation of the data.

I attach an infographic to further explain conceptually what I am talking about! (courtesy of my friend John Zorich, MS, CQE, who developed his own Excel spreadsheet version of this method, with many textbook references).

JMP has the "Life Distribution" Platform, and "Custom Estimation" and these tools provide some semblance of the Reliability Plotting method, but bottom line is they are not designed to output the C/R in a direct, intuitive, and graphically-driven way.  I would love to know if anyone else is familiar with this technique and if there others would find it useful to have a ready implementation in JMP?

Thanks! I include @danielvalente at JMP whom I've discussed this with in recent weeks.

Cheers, @PatrickGiuliano