JMP Wish List

There are at least 9 different ways to calculate quantiles. JMP only supports one of them (Weibull estimate). Requesting that other options be supported and user choice allowed for selecting which option is to be invoked. Other SAS tools appear to support multiple options. Based on a 1996 paper (Hyndman and Fan, "Sample Quantiles in Statistical Packages", The American Statistician, Nov. 1996), the Unbiased Median approach is considered to be the most robust. Here are some relevant links that provide further details.   https://mathworld.wolfram.com/Quantile.html https://en.wikipedia.org/wiki/Quantile_function https://analyse-it.com/blog/2013/2/quantiles-percentiles-why-so-many-ways-to-calculate-them   Thanks and regards Mehul Shroff ... View more

Text explorer is a very exciting platform, however it doesn't seem to have the same interactivity than other JMP tools and I really missed it when creating dashboards. It would be great if the word cloud would react and change depending on the selection on the other graphs. At the moment you can only right click on a word and select rows. This allows you to see the selected rows in the other graphs, but it doesn't work the other way round. On top of that, this possibility disappears when converting the dashboard into an interactive HTML. ... View more

It would be helpful to be able to change the default # of levels for the group X binning as well as be able to save out the current binning from the graph to a new column (similar to the name selection in column). ... View more

Can we have JMP16 correct standard errors for non-normality (e.g., Satorra-Bentler correction) or use bootstrap for estimating variability and goodness of fit measures (e.g., Bollen-Stine bootstrap) in the multivariate SEM platform? ... View more

I'd like to see the ability to sort the X axis in an ascending or descending order based upon one of the levels in the overlay box.  So when your right click on the X axis there would be a new selection item that would default to "all" but would list the levels from the overlay box  and you could select which one to use.  You'd then do your normal "order by" choice. ... View more

Issue: I am trying to stack together different analyses into a single profiler table. However, Some researchers have used different factor ranges in their experiments. When I stack the profilers together JMP uses the largest range across all the data tables for the factor settings on the bottom of the profiler. As a result, some analyses are making predictions outside the bounds of their respective experiments leading to erroneous predictions. Suggestion: It would be nice if JMP understood the factor ranges of the respective data sets which generated them so that when I combined the profilers it gives a null prediction or somehow grays out that region of the factor space that wasn’t included in the respective data table. I know there is an extrapolation warning feature being added to JMP 16 but I think it would be better to either null out the prediction or gray-out that extrapolation space.  ... View more

Issue: Sometimes categoric factor settings have long names and it makes sense to change the alignment to either vertical or horizontal in the axis settings so you can see the whole name. However, when you do this on the scatterplot matrix it changes both sides of the factor settings on the x and y axes. So I can get it to look right on one axis but not the other. Suggestion: Decouple the factor axes so I can make a change to one axis while not affecting the other. ... View more

Suggestion: Build in confidence and prediction bounds in the Surface and Contour profilers and allow the user to toggle them off & on and to set the alpha values in a pane off to the side. You can use a different surface transparency on the surface profiler so you can still see the underlying shape of the prediction curve. Would be great if the confidence and prediction intervals could be integrated into the projected contour plot underneath the surface profiler and connected to the Response Grid Slider. This is super useful in explaining how a response requirement i.e. the Grid Slider intersects the prediction surface. Would be super useful in explaining to project management when "it depends" whether or not a requirement is met. ... View more

It would be very nice if Evaluate Design allowed you to use a local data filter when evaluate a single design (single table).  The use cases are:   Evaluating the impact of lost design point(s) Quickly seeing evaluation of intiial and then subsequent augmented designs.   I'm not suggesting you should be able to compare designs this way, this is only for rapid check of design properties for subsets of a full design table.   ... View more

I would like the interactive html reports to support the surface profiler and/or 3D contour profiler. This would be very helpful for sharing results from a response surface model that are not represented well by the 2D profiler.  ... View more

I recommend adding a table of p values for Ni+1 to the Pareto Plot. This would help in statistically determining which categories in the plot to address. The details and calculations are shown in Quality Progress Nov 2020 issue pages 38-43. The article is 'Critical values for Pareto comparisons remove statistical subjectivity' by Frank Rudisill and Jim O'Connor. The calculations are on page 42. Table 3 on page 43 shows an example containing a p value for Ni+1 which could be added in part or in full to the Pareto Plot. ... View more

In capability analysis often only the upper part of the observed distribution is relevant. So, why not add an option to fit just the upper part? This will make the fit often much better in describing the part that matters, using e.g. a half normal distribution. ... View more

In graph builder many variables can be added to both x and y axes to make a scatterplot matrix, but often the user only wants the bottom left triangle, as displayed in the scatterplot matrix platform.  Provide an easy way to remove all of the diagonal and top right plots (or at least the data in them) so a user can recreate a scatterplot matrix but with all of the options and customizability of the graph builder platform. ... View more

The Multivariate platform does only accept one input list. Once the dialog is run, correlations will be calculated on all pairwise X from the input list. Similarly when adding a Scatterplot to the multivariate analysis, all pairwise scatterplots will be shown. However, in many situations there is an interest in only "correlating" and creating a scatterplot on a certain list of X variables versus another list of X (or Y) variables in terms of exploratory data analysis. Since all variables are always there this is a tedious task to scroll through that massive matrix of scatterplots in scenarios of wide datasets. I'd appreciate to see two input lists for Multivariate with pairwise correlations/scatterplots from the variables in one list versus the variables in the other list. ... View more

Dear JMP Team,   I'd love to have the ability to generate a summary table where the data itself was linked to the source table. And if I'm being greedy, I'd like to be able to use the summary table to generate a table of stacked columns which also link back to the original source table. This would ultimately result in the ability to change the source data and have the change reflected in a stacked summary table.   Thanks,   Kevin Herlihy ... View more

I wish different lines on a x-y plot in Graph builder would appear with different symbols in addition to different colors, by default.  The reason by default is important is because, as someone who is slightly (very?) color blind, I find it difficult to differentiate between the different overlay categories simply based on color.  Often I'll manually change symbols in the dataset one by one in order to see the differences. ... View more

As a clinician researcher in translational science, I love JMPs ease of use and powerful features, visualization, etc. But there are features that would make it more useful for multi-omics data: 1) support/help/guidance/platforms that focus on small sample sizes with lots of data (i.e. your typical medical basic research where the number of patients is limited but the number of variables per patient is large). A commonly used algorithm that is missing is OPLS-DA, which granted is just an application of PLS-DA which JMP has. You do have lasso which is also popular. Implementing other neural network/machine learning algorithms would be great (XGBoost!). But the point is to integrate the different options into an add-in or platform specific for these tasks, and the guidance/help associated. 2) Support for longitudinal sampling: many times the patients are sampled more than once. Adding that to the above would make JMP stand out and lead the field for these types of analyses. 3) variable selection is crucial, and using different ways is important to see how they perform, i.e. forward addition or backward elimination, bootstrapping (to maintain the #of predictors per variable) vs sub-sampling, etc. Having a profiler for this type of analyses would be amazing, or having a learning curve simulation that shows how the models are expected to perform for given parameters used. 4) The associated visualizations and model comparison and choosing.  5) Another very important feature is network analysis. I don't mean the neural/ML versions, but the ones that use correlations between different variables to create their maps of associations. It's like doing the SEM in the reverse direction. This would be especially useful if done at the individual level i.e. take every patient's scatterplot matrix, and look at the same pairs accross all patients to see which variables are consistently correlated. This is different than putting all the point from all the patients together in one big sample.     Perhaps some of these features exist in separate parts of JMP, and I guess some could be automated/recreated using JSL. But for those of us who are drawn to JMP because it is not R, then having these features organized coherently like other JMP platforms would make a big difference. I can also say that this is the same problem with all the other leading commercial stats packages, so a robust implementation in JMP would be a differentiator.  ... View more

I noticed Sall's presentation that when he held down the command key to apply changes across (eg the histograms) that the text of the key stayed the same. The Apple User Interface Guide would have the text change to provide user feedback. ... View more

It'd be cool if we can right-click on 1 point/column of scatter/bar chart and display its value as a label on the graph builder. Sometimes, we only want that value highlighted, not all crowded labels :) ... View more