Notes/General
1. Auto-hide menus and toolbars
Data Import, Data Table, Data Visualization, General
2. Compare Tables refresh Improve Compare Data Tables feature to show the data (rather than just select rows) and make the comparison more flexible.
3. Make Binning Formula Dialog Improvements Incorporate interactive binning add-in functionality into JMP.
4. Standalone Regex Dialog Currently, the customize Regex dialog is only available through Text Explorer. There are features of this dialog which are useful outside the context of Text Explorer. Craige had added a button to interactively save the result of the text parse to a new column but it was still tied to the Text Explorer flow. This feature will make the Regex dialog available directly from the Cols->Utilities menu.
5. Hex binning on Graph Builder Heatmap Heatmaps currently bin in rectangular bins. This feature will add the ability to bin in hexagonal shapes.
6. Size role support in Heatmap Currently, the size variable is ignored by the heatmap. It can be used to scale the size of the bin and should be supported.
7. Data Filter - new unstructured text filtering Allowing unstructured text columns in the local and global data filters. Makes it easy to find key words, filter on key words, or create subsets based on those key words.
8. Custom date formatting Custom date formatting in the form of Format Patterns. It allows the user to specify the date format as a string such as “ <:><:>”. The text delimited by angle brackets are called field descriptors and are replaced by the appropriate value (“” would be replaced by a four-digit year). These formats can be used for both formatting and inputting data.
Design of Experiments
9. Sample Size Explorer Extends the current power and sample size calculator to be more interactive and visual. Encourages exploration of the interplay between sample size and power/interval size/demonstration time.
10. Bayesian D-Optimal Design Fast construction of 2-level D-optimal supersaturated designs.
11. Bayesian D-Optimal Design Enhanced handling of covariates in Custom Design. Functionality will be further improved in EA 3.
Statistics/Modeling
12. Model Screening A platform to run all the prediction modeling platforms to help pick the best-performing.
13. Support Vector Machines - Regression Extends the Support Vector Machine platform to do regression in addition to classification.
14. SEM Multiple Group Multiple-group analysis enables investigation of how effects vary across groups, which are defined by a categorical variable (usually with few levels).
15. SEM improvements SEM Profiler. Path diagram. Model Shortcuts. Model Comparison table.
16. Sentiment Analysis Add a simple form of sentiment analysis to the Text Explorer platform.
17. Curve DOE in Fit Curve Analyze how supplementary variables impact the parameter estimates from the nonlinear regression model. We can do this in the Fit Curve platform.
18. Categorical, FDR Adjusted PValues Categorical has a new option under “Test Options” submenu, “FDR Adjusted PValues” which will add columns to most of the report tables that have p-values showing the False-Discovery-Rate-Adjusted PValues (Benjamini-Hochberg). This is useful when you do many tests and want to protect against tests being significant by random chance alone.
19. Annimation for Profilers A button will start an animation tour of changing the factors, one at a time to watch how the profiles change as you change each of the factors.
20. Time Series Forecast improvements The new Time Series Forecast platform enables you to forecast multiple time series, where the number of series could be very large (hundreds). The platform provides compact summaries of the multiple time series and provides options for efficiently fitting up to 30 different forecasting models in the exponential smoothing family. The best fitting model is automatically selected. The Time Series Forecast platform is designed to select a best fitting model automatically from a set of exponential smoothing models based on the Hyndman et. al. state space smoothing approach.
Quality Engineering/Reliability
21. EWMA Chart Platform Adding a new platform which combines EWMA with SPC and Feedback Adjustment, from Engineering Process Control methodologies.
22. Control Chart Builder customizations Right-click within the Control Chart builder to easily remove the dispersion chart, or to hide or display particular control limits or the center line. Use a top red triangle menu option to show limit labels on the control chart.
23. Reliability improvements MTTF calculation in Fit Life by X. Support Box-Cox scale in axis in Fit Life by X. Add residual versus fitted in Fit Life by X diagnostics
Output/Sharing/Scripting
24. Project improvements Container Projects. Project Preferences. Opening Images. Data Filter Tool Window.
25. Error List Feature Improve visibility and usefulness of errors, warnings, notes, and other information given to the user by JMP. And improve the quality and quantity of logged information.
26. Column Switcher support for Dashboard Builder The initial integration of the new shared Column Switcher has been pushed to Dashboard Builder.
27. Set script to run after column switch This is a JSL only option. An example is below. We currently don’t write out the script when writing a platform back to JSL. It wouldn’t work if publishing to JMP Live. It would be interesting to get feedback on that limitation.
28. Multi-platform Column Switcher This is another JSL only option. There is probably a limited audience for this feature. Not sure if any of the EA customers would be interested. JSL example is below. We don’t support this yet in Dashboard Builder but Dan S is working on it. We may be able to support it in JMP Live from a dashboard once we work that out. The nice thing about this implementation is that the column switcher and platforms can be in different windows and there can be multiple switchers switching multiple platforms.
Other
29. Xgboost XGBoost is an experimental platform interfacing to the XGBoost program
Not on Single Source Document List
When data tables are small, using a training set that is a subset of the data is problematic, making crossvalidation hard. Sometimes things are even non-estimable. However a weighting method is an alternative. It involves making a duplicate of the data appended to the original data, with anti-correlated weights between the training and validation copies of the data.
- Avoid Extrapolation in the Prediction Profiler
Provide an option to prevent (or alert) the user from extrapolating too far outside the range of the original data when exploring predictions in the profiler or performing optimization.
- Covariate handling in DOE
Enhance handling of covariates in Custom Design.
The objective of this is to infer specification limits on the factors that insure compliance with specification limits on the responses. This is a fundamental part of the Quality-by-Design initiative of the FDA in the pharmaceutical industry.
- Enhancements to Transform Columns (4 enhancements)
- Nesting - allow a transform column formula to reference another transform column.
- Table Scope - ability to create transform columns at the table level .
- Caching - cache transform data at once .
- Virtuality - allow normal property set and get and be usable as much like a real column as possible.
- Explore Outliers for JMP16
For Univariate Outliers, more commands to make new columns with the outliers converted to missing values. For Multivariate Outliers, three new methods: DDC (Rosseau) hdOutliers (Wilkinson) and Robust PCA (Candes). All these will be better and faster than the Robust Correlation outliers that we call in the Multivariate platform. We also have improved the k-nearest neighbor internals to be much faster. LaFarge
Functional PCA model selection / diagnostics. Model free Functional PCA for large data. Function ML / Functional Input modeling options. Interface for GenReg models fit in FDOE. Save Formula options. Reduce for FDE Group. Summaries for data processing steps . Data processing details. Fused Lasso modeling option .
- Functions For JMP16
- Some new "programming" functions for working with containers in JSL. These include "For Each" and "Transform Each".
- GraphletFootprintReduction
The primary goal of this feature is to reduce the amount of JSL used in the definition of Graphlet Presets. This could be accomplished by relying on high-level JSL functions defined in a library or package that would loaded (included) at runtime. Which means we now have to find a way to 1) ship JSL code and 2) make sure this code can be found by the graphlet engine.
Improve useability for the visually impaired.
Improve useability for the visually impaired.
- HoverLabelEditorImprovements
Make hover labels easier to generate.
Improved support for image files. New features: Panning and Zoomng. Ordered Axes. More interactive reports.
Make Journals easier to use.
Fixing two big issues: (1) ignoring blocks if unbalanced, (2) inappropriate statistical methods if balanced blocks used.
- Partial Correlation Node Diagrams
A graphical depiction of correlations where the points for variables are close together are highly correlated and connected by lines (command in multivariate)
- Profiler Optimization that will observe disallowed combinations
Add an optimization algorithm that can handle the disallowed combination constraints that can be created in DOE.
The sequence of columns selected in a drill-down is called a hierarchy. Column Chains (per column links). Column Lists (supports multiple hierarchies per table).
- Missing Value in Gradient Legend
Each gradient defined in JMP will now also include a missing value color.
Smoothers for JMP 16
Developer & Early Adopter Videos
