JMP Wish List

  ☑ cool new feature ☑ could help many users! ☑ users can remove bugs on their on ☐ nice to have ☐ nobody needs it   What inspired this wish list request?  Via  a user can edit the main menu and [on windows systems] the toolbars.   Other menus cannot be edited. Unfortunately, some of the implemented functions have bugs - and some functions are missing.   What is the improvement you would like to see?  Please add the possibility to remove items from any right click /  red triangle / ...  menu and the possibility to add user defined items. This will allow us to fix tiny issues which won't make it into a future version of JMP - or just in many years. e.g. - Allow us to perform more actions when right clicking on column name on columns panel  - Paste Column has a bug and should be replaced with a version without bug: JMP jumps   Is it possible to customize the Add Formula Column menu to include customize functions?    yes and no: -  a user can not add  user-Defined mini GUIs , e.g. to add Modulo to the portfolio. - Existing functions cannot be added to the menu by name.   (TS-00152405)     Why is this idea important?  Less pressure on JMP developers to bugfix and fine-tune the GUI. Users don't have to wait for version N+2. -> Users will be able to bugfix and tweak the GUI on their own.   other wishes from  ... View more

What inspired this wish list request?  @FN linked to a scikit-learn article in a comment on theValidation for Continuous Processing Data add-in page.  This is a better way to do cross validation with continuous process data and it reminded me that the techniques used in that add-in would be much more effective and used more often if they were built into the JMP. This situation applies to many analyses my colleagues and I do regularly with manufacturing data.   What is the improvement you would like to see?  - Incorporate grouping by time in to the made a validation column dialog box, including similar functionality to what is in the Add-in linked above to guide the user to an appropriate group size. - Add an option for an individual table so crossvalidation used everywhere in JMP uses a time-based method referenced in the link below instead of randomly assigning rows.   Why is this idea important? Although @DrewLuebe and I created the validation add-in linked above specifically to help users better understand the predictability of models in data sets that have correlation between rows, including autocorrelation, the built-in crossvalidation behavior and default validation still use per-row splits.  This means that many reported validation fit metrics are artificially better than the actual performance will be with new data .  By baking these techniques into JMP itself, users will have a much better understanding of their data and models no matter what tools they use.     ... View more

What inspired this wish list request?  Model selection is essential for DoE Analysis. JMP provides this feature for platforms like Fit Stepwise or Generalized Regression, where only fixed effects are present in the model. However, by using Split Plot Designs or random block models I need to use either the Mixed Model Platform or the Generalized Mixed Model Platform. Both platforms does not allow automatic model selection steps like Backward Elimination, Forward Selection or Prune Forward Selection. Thus, I'm forced to perform these steps in a manually way, which are very tedious. Other software tools like Design Expert already allow automatic model selection for split plot designs!     What is the improvement you would like to see? Simply add all the model selection features from the generalized regression platform (see screenshot) to the Generalized Mixed Model Platform       Why is this idea important? At the moment I have to use Design Expert to perform a model selection on Split-Plot Designs, because it is too much work to do this manually in JMP. In other words, I cannot use JMP for Split-Plot Model Analysis respectively not even for random block models!     ... View more

1. Multiple-comparisons for nominal Y's (as for continuous Y's in Fit Y by X and in Fit Model). 2. Multi-way contingency tables for factorial layout of X's, and a nominal Y. 3. Contingency table: give proper titles to the last column and last row (e.g. "Y sub-total" and "X sub-total"). 4. Contingency table: when "Col%" is requested, it should be presented also in the "Y sub-total" column and when "Row%" is requested, it should be presented in the row of "X sub-total". ... View more

1) I would like to suggest adding the option to bootstrap to the validation process in the platforms. Either as a validation option directly (i.e. as one more alternative to the existing K-fold, holdout, etc), or as a nested option i.e.bootstrap the holdout or K-folds. In my mind it is more intuitive and makes sense from a workflow perspective. Also, if one bootstraps at the end by  right-clicking results, one gets the bootstrap only of a specific table and not the whole model's estimates i.e. if I bootstrap a table, I get the table results but not the results from the residuals or diagnostics etc. Moreover, the resulting table is raw and unprocessed and lacks context; it is also more involved to use those results to assess and improve the model. If I bootstrap  a model I would then need to take those results and further work with them to integrate the results into my analytical process and decision making. Having the bootstrap integrated into the modeling platform would include the bootstrapped estimates and results for the whole model as separate table results. It would leverage the bootstrapping option to really use it for improving modeling (better models, more analytics on their performance and stability, etc).   2) Related to that is adding the option to bootstrap every model fit in the platform. For example logistic fits don't have validation, and using the bootstrap with the output table is very onerous to the point of being useless for a user that doesn't know scripting and stats theory to high level. Adding the option to bootstrap to the platform would make it much easier to manage. Nowadays doing validation is standard for almost any fit where there is a significant number of predictors, regardless of the fit platform used.   Uriel. ... View more

What inspired this wish list request?  JMP has made a range of machine learning algorithms readily available - a major deficiency in most of these is the lack of uncertainty measures for predictions, particularly for classification models.  There are a number of tools available for assessing model accuracy (e.g., ROC curves, confusion matrices, etc.) but few for assessing the uncertainty in the predictions.  Unlike classical statistical models (where there is an assumed error structure), the degree of uncertainty in the predictions is often lacking.  Conformal prediction appears to be a relatively straightforward way to provide this for a variety of machine learning methods.     What is the improvement you would like to see?  I would like to see the conformal prediction capability attached to any of the classification modeling methods employed in JMP.  The resulting output should show (for classification models) the predicted classification sets, including an indeterminate set where the desired level of certainty cannot be assured.      Why is this idea important?  Conformal prediction has been around for at least 20 years, but is hindered (in my opinion) by an unnecessarily mathematical description of the technique.  I think it is more straightforward than as typically presented.  I also think it is a nonparametric technique for quantifying uncertainty in predictions - sorely needed as machine learning models are increasingly used.  The lack of quantifying predictive uncertainty (rather than just accuracy) limits the usefulness of these methods, and potentially makes their application misguided (yielding predictions where there may be little to distinguish between observations that can reliably be predicted from those for which it cannot.  For one reference, see https://arxiv.org/pdf/2107.07511.pdf.     ... View more

What inspired this wish list request? @scott_allen has developed an add-in that allows tuning of NN parameters,     What is the improvement you would like to see? Adding more layers to the NN platform will allow faster and better model building and exploration using the NN-add-in.     Why is this idea important? NN models are subject to a lot of parameter tuning and a transparent and efficient DOE based approach has a lot of advantages in robustness and also accuracy   ... View more

What inspired this wish list request?   In JMP version 17, it is quite time consuming to randomly fit numerous neural net models to hopefully find one that fits well.  If we don't type in the seed number at the start we cannot reproduce the exact analysis later, so we cannot currently reproduce the fit versus validation plots and prediction profilers.     What is the improvement you would like to see?            Why is this idea important? Development of neural net models in JMP is a potentially powerful feature, but it needs improvement to make the modeler's job a bit less repetitive in searching for a neural net that well fits the data.       ... View more

I would like to propose the ability to have scriptable control for code folding markers. This in my opinion would enable easier management of code. For e.g.  SelectableMatrix = Expr( ResMat = {}; // loop through each row For( i = 1, i <= nR, i++, ResMat[i] = {}; // loop through each column getting values For( r = 1, r <= nC, r++, ResMat[i][r] = TB1[CheckBoxBox( r )] << get( i ) ); ); ResMat = Matrix( ResMat ); ); The ability to then use actions like the following to contract and expand the code will make it easier to manage the code.    SelectableMatrix << expand; SelectableMatrix << contract; ... View more

I would like to see the RSquare validation value available for each row in the output table with the all possible models tool.  I find myself using the all possible models option often but I always end up needing to manaually create the exact table to include the validation information.  The information is already avaialbe without actually running the model in the stepwise regression control on the top of the page.       ... View more

I know it is possible to do bootstrapped CI in JMP. But, when we build and test complicated mediation model, the indirect effects need to be calculated and then a bootstrapped CI is needed to build on that indirect effect. I am sure some smart JMP user has figured it out. I hope that person may write a script or a step-by-step procedure for doing this in JMP. Otherwise JMP developers, please take a note of this wish list. ... View more

Really would like more than the current hidden layers available.  JMP Pro makes analysis very easy for CART, Random Forests, Naive Bayes, and SVMs but the Neural Net model is limited to too few layers.  Consequently I go to R to build more sophisticated deep nets.  Would be so much easier if this capability was incorporated into JMP.  Thanks, Shaun ... View more

In continuation of this exchange: Interactions in OLS regression and simple slopes.    I would like to propose that JMP in a similar manner to SPSS Process or simple R scripts would include in multiple regressions that calculate interactions between variables, the independent value of each variable slope and not only their additive value.  ... View more

Since JMP emphasizes visualization, the augmented factor correlation matrix with color coding will be a useful piece of information in the SEM platform. The factor correlation matrix is a useful for researchers during scale construction. For multi-item constructs scale builders look for convergent and discriminant validity. This piece of information can be obtained easily since JMP 15 can now compute factor scores for latent variables. This correlation matrix can be augmented as follows:   1. Correlation values on the lower diagonal 2. Correlation-squared values reflecting shared variance between factors on the upper diagonal 3. Average Variance Extracted (AVE) which is the variance each factor shares with its own indicators on the diagonals. This can also be thought of as the average communality of the construct.   Here is an example:   For convergent validity, we want AVE > 0.5 indicating the construct explains > 50% of the variance of its indicators   For discriminant validity, we want the diagonal values to be higher than values to the right or above since this establishes that the factor shares more variance with its own indicators than other factors. When such is not the case, color coding the violations will be more useful. ... View more

I suggest that the output for regression include options for the Johnson-Neyman regions of significance—i.e., floodlight analysis (i.e., at and above/below what value of the moderator is the difference between levels of the independent variable significant). In addition, I suggest including options for spotlight ananyses, based on regions defined by the user (e.g., significance at +-1 SD from the mean of the moderator or at a particular value of the moderator).      ... View more

What inspired this wish list request?  I am analyzing data for an engineer in which part of the data is below the detection limit, which is good. I want to include that data in the results. The data is a batch process, and the batch should be included in the model as a random term.   I tried using GenReg (which accepts interval censored data), but it would not allow the random term in the model. I tried using Standard Least Squares, but it would not acknowledge the interval censored data. I tried using Generalized Linear Mixed Model, but it would not acknowledge the interval censored data.   What is the improvement you would like to see?  I would like a modeling platform that allows random terms in the model, and interval censored responses.   Why is this idea important?  This would allow me to build models for my engineers that do not have glaring problems or missing data.   ... View more

What inspired this wish list request? I have found functional data explorer to be useful for modeling kinetics. It would be really nice if I could share the prediction profiler with a wider audience.   What is the improvement you would like to see? I would like to see the ability to view profilers for platforms like Functional Data Explorer available in JMP Live.   Why is this idea important? It would allow me to show more advanced modeling prediction profilers. These are useful for dialing in recipes with margin differences.     ... View more

Thanks to Russ Wolfinger and JMP Super-Users like @SDF1, there are now add-ins that allow easy tuning of hyperparameters for XGBoost, neural nets, bootstrap forests, and others, but I would really like to see these integrated into JMP. One thing I would like to do is test out thousands of neural networks with multiple Y's, which cannot yet be done with the add-in (and is not possible at all for XGBoost, as far as I can tell).      Sometimes, we need to predict more than one thing. For instance, maybe you want to optimize a process while minimizing costs. With several JMP modeling platforms, you can create a two-Y model, then use the desirability analysis under the profiler to do this, but you cannot build thousands of dual-Y models using neural networks, bootstrap forests, or even standard least squares. I think this would really increase the capacity of JMP Pro to do some hardcore predictive modeling.      ... View more

What inspired this wish list request?  Creation of validation columns in JMP and having to add the preselect role myself (or just not utilize it) even though the column has clear preselect role.   What is the improvement you would like to see?  I would like to have an option in Make Validation Column platform (or make it always happen?) to set my Validation column Preselect role as Validation.   Why is this idea important?  Makes use of usually not very used Preselect Role and makes it a bit faster to use Validation columns in platforms which supports it.     ... View more