New in JMP^® 14 and JMP^® Pro 14

Projects afford a new way to organize files and use JMP with a tabbed, web browser-like interface and can ease the pain of managing your workspace when many tables, scripts and reports are open at once.

The single-document interface provides a place to bookmark important files and displays a window list to let you see and easily move to any open window, data table or script. If you run repeated data queries, analyses and reports, using a Project is a quick way to collect a set of workflow tasks for streamlined sharing, saving and collaborating. With Projects, you can:

• Choose which files to include in the project.
• Launch supporting documents (.ppt, .doc and .pdf) directly from JMP.
• Share and archive analyses.

For many users, each analysis begins with importing dozens, hundreds or even thousands of files. Before JMP 14, these users needed to write JSL scripts to perform such imports. With JMP 14, the Multiple File Import platform gives users a point-and-click way to quickly and accurately combine hundreds— or even thousands—of files into a single one. 

Import multiple files into a single data table, or a collection of data tables (for example many free-text documents that need to be concatenated for use with Text Explorer), with just a few clicks. Specify import options, filter files by a variety of criteria (including size, date, name and type), and import to a single column (for use with Text Explorer) or into a multi-column table. Use Multiple File Import interactively, or have the platform write you a script to automate future imports.

Graph Builder is the graph creation toolbox in JMP that helps you discover and share the stories within your data. With JMP 14, Graph Builder has new capabilities that enhance the data exploration process and further the publication-readiness of your findings. Among these new features is a new bar type that combines elements from treemaps and Pareto plots and many new options that give you finer control over graph customizations. Graph Builder receives many other enhancements in JMP 14, including:

• Easily incorporated customized error bars.
• Marker Sizes from legend.
• Point dodging and jittering options.
• More control over contours including smoothing and shape clipping.
• Bootstrap confidence intervals for Smoother element.
• Support for additional basic statistical analysis directly in Graph Builder. 

With direct database access and an ODBC driver, it is easy to query, join and filter data using Query Builder—but what happens when you'd like to perform an analysis using one of the many data sources available through web APIs?

JMP 14 introduces the JSL HTTP Request function. HTTP Request allows you to write custom scripts to communicate with external web servers (for example, Salesforce, Google Analytics or Twitter) through REST APIs. By using HTTP Request and the new JSON parsing functions, you can automate the process of fetching data from the web and placing it into JMP tables.

Every JMP user works with data tables, and in JMP 14, they are even more powerful and easier to use. Now tables can use the values of any column—even an image column—as markers. They can also store hyperlinks to web pages, files and JSL scripts and perform several new single-click transforms. Additionally, more powerful Virtual Joins are possible with just a few clicks and a new column utility can easily convert between value labels and codes. A few other new features to enhance your use of tables include:

• Improvements to copy-and-paste functions that let you paste free-text into a single data table cell.
• Grouping and nesting of table scripts for better script organization.
• Improved automatic category sorting for numeric suffixes (e.g., Lot 1, Lot 2...Lot 10, Lot 11).
• New invert column selection option.
• Ability to easily select duplicate rows from the rows menu.
• Updated Scripting Index with improved search capability, and a compact mode that pairs well with Projects.

Ensuring the consistency and accuracy of categorical data has always been one of the most time-consuming and tedious aspects of data cleanup. Recode is a tool that many scientists and engineers utilize daily to expedite the data cleanup process. In JMP 14, you'll find that Recode has additional tools and automated routines to get you analysis-ready more quickly than ever before.

With improved filtering and text parsing options, support for multiple response data and value labels and integration with the Formula Depot in JMP Pro, you won’t trust your data cleansing to any other tool. Here are just a few of the Recode enhancements in JMP 14:

• RegEx support.
• Import Recode results column.
• Ability to create, consume and convert value labels directly from Recode.
• Option to publish recode formula to Formula Depot, which lets you perform data cleanup inline in a production environment without having to bring the data into JMP.
• Support for multiple response columns.
• Ability to group to new category.

JMP 14 introduces Python into the suite of programming language interfaces in JMP, adding to the existing ability JMP has to connect to SAS, MATLAB and R. With the Python interface, JMP users can leverage in-house Python experts and pre-existing libraries, without leaving JMP. It also allows users to connect to a local install of Python, send data from JMP to Python, execute Python code from a JSL script and return data to JMP for data visualization, analysis and further exploration.

The DOE platform in JMP is world-class, better enabling users to solve challenging problems in a wide variety of real-world settings. In JMP 14, numerous improvements to the DOE platform include:

• A-optimality criterion added to Custom Designer. This is useful if the optimal design needs to focus more heavily on certain effects rather than others.
• Ability to create Balanced Incomplete Block Designs (BIBDs).
• Improvements to the way priors are specified when building Accelerated Life Test (ALT) Designs.
• In Compare Design, up to 10 designs can now be compared through a JSL interface and up to five with the UI.
• Improvements to the Fit DSD algorithm and options enabling the platform to find weakly quadratic terms and interactions having weak heredity.
• Addition of a pairwise incidence heatmap to MaxDiff designs.
• Improved space-filling over levels of a categorical factor.

JMP Pro 14 brings improvements to the Naïve Bayes functionality introduced in JMP Pro 13. This classic approach allows classification of categorical responses in the data and works well with large data sets. It uses a simple estimate of the probability of conditional distributions of X given Y and reverses the conditional probability using Bayes' rule. In JMP Pro 14, Naïve Bayes is rendered as a fully featured platform and offers the functionality available in other predictive modeling platforms, such as profilers and diagnostic plots.

As with the Naive Bayes platform, JMP Pro 14 improves upon its K-Nearest Neighbors (K-NN) functionality. In JMP Pro 14, K-NN is now a complete platform with features available in other predictive modeling platforms, including profilers, a model selection tool that lets you interactively select K and diagnostic plots to help you build more robust models.

Text Explorer was an exciting addition to JMP 13, letting users explore unstructured text data in a point-and-click environment. Now, Text Explorer has enhancements that make it even more useful for digging deeper into text. Some of the updates available in JMP 14 include:

• Chinese and Japanese language support.
• Better filtering options for tabular output, including terms, phrases and report lists.
• Per topic word clouds.
• Support for discriminant analysis (JMP Pro).
• Support for validation column in the platform launch for use in a predictive modeling workflow (JMP Pro).

Quality engineers will find a variety of new diagnostics and time-saving tools in JMP 14. CUSUM charts in JMP now have the familiar look, feel and interpretation of traditional control charts and the Process Screening platform adds Drift Detection, Goal Plots and the ability to generate three-way control charts directly from platform launch. The Process Capability platform adds new distributions and measures of variability, while a new platform, Manage Spec Limits, lets users easily import, export and modify spec limits.

Multiple Factor Analysis (MFA) is a sensory analysis technique whose goals are to find groupings of products that are similar and to identify outlier panelists—the panelists who are so different from the rest of the group that their values skew the results of the analysis. MFA allows the researcher to perform a PCA-like analysis with untrained panelists. It is often used for pre-evaluation studies, where many different products are being tested, as well as for later-stage consumer panel studies where many consumers—many of whom have no training on standard sensory measurements—rate the products.

Scientists and engineers are dealing with more and more data streamed from device-based sensors or batch process monitors. Hundreds or thousands of such data streams can create a large volume of data, very quickly, posing a unique set of challenges.

No matter what your industry, common difficulties exist with this type of data: simplifying and cleaning up messy data, removing outliers and building models that characterize an underlying function or relate a continuous data stream to measures of performance, such as yield, defect rate or product quality.

With Functional Data Explorer in JMP Pro 14, you can take advantage of this functional data, easily performing data cleanup, transformation, visualization and feature extraction. Ultimately, this allows you to more easily build models you can use to design experiments, characterize processes or make predictions.

Generalized regression is the premier linear modeling platform in JMP Pro. With Generalized Regression, you can build models for many kinds of data—for example, DOE data, observational data, data with categorical responses, messy data, text data, highly correlated data and more. It is a tool that fits models, selects variables, handles multicollinearity and provides diagnostic reports—all in one place. JMP Pro 14 introduces some key enhancements to the Generalized Regression functionality, including:

• Support for censored response data now includes the Gamma and exponential distributions.
• Multinomial response support (nominal responses with more than three levels).
• Additional selection techniques: best subset regression, backwards elimination, pruned forward selection and the Dantzig selection (which has gained popularity in the DOE community for analyzing n<<p problems, liked supersaturated designs).
• New reporting features include a reset solution button, QQ-Plots, F-tests and reporting VIFs, adjusted R-squares and RMSEs.
• The platform has been adjusted for improved performance with large and/or complex problems with many effects.
• Generalized Regression now supports custom tests, and cross-validation within two-stage forward selection.

• The new Projects system provides a way to organize files and use JMP with a tabbed interface to manage multiple open windows during analysis. Projects allow single document interface to JMP, selection of specific files to include, organization of a reconfigurable workspace, analysis sharing and archiving and document launching (of supported file types PPT, DOC, PDF) directly from JMP. 
• New multiple file import launches a dialog to load several files into a single JMP data table. Allows filtering of files by size, date, name and type and supports one-column import, useful for Text Explorer document preparation. Within platform support for editing text import options.
• Virtual Join improvements offer access to data in one table by another table with just a few clicks. 
• Graph Builder improvements include:
  • New chart type, packed bar chart, shows skewed data across hundreds of categories.
  • More control over graph customization.
  • Easily incorporated customized error bars.
  • Marker sizes from legend.
  • Support for more basic statistical analysis directly in Graph Builder.
  • More control over contours including smoothing and shape clipping.
  • Bootstrap confidence intervals for Smoother element.
  • Point dodging and jittering options.
  • Parallel coordinates in box plots, points, histogram and violin plots.
• New data table features include:
  • Use column images as markers. 
  • Paste into cell.
  • Grouping and nesting of table scripts for better script organization.
  • Clickable links afforded by the Event Handler column property.
  • Invert Column Selection option.
  • New transforms are now in New Formula Column.
  • Labels to codes and codes to labels columns utility.
  • Select duplicate rows added to rows menu.

• Interface enhancements improve workflow:
  • Preference searching and filtering makes it easy to find a preference to be changed.
  • Updated Scripting Index with better search and a compact mode that works well in a project.
  • Preview results in Formula Editor before clicking OK.
  • Better automatic category sorting for numeric suffixes (e.g., Lot 1, Lot 2… Lot 10, Lot 11).
  • Grouping of scripts in a data table.
  • File quarantine alert to protect against unknown executable code within downloaded JMP files.
  • Updated OpenGL View Settings dialog for easier customization any 3D graph.
• Enhancements to the Recode feature:
  • Search box for more control over filtering.
  • Improved Text Parsing.
  • RegEx support.
  • Recode results column import.
  • Ability to create, consume and convert value labels directly from Recode.
  • Support for multiple response columns.
  • Recode formula publish to Formula Depot, which lets you perform data cleanup inline in a production environment without having to bring the data into JMP.
  • Group to new category.
  • Recode numeric continuous columns.

• Animated GIF recording and saving in JMP where there is a play button (Bubble plot, local data filter, column switcher for example) make it possible to embed a moving JMP visualization in a PPT slide, for example, without having to rely on Flash.
• Dashboard Builder adds easier-to-read drop zones, a blank dashboard template and the ability to store custom templates.
• Now select data in HTML5 report output.
  

• Create Web Report functionality allows organization and sharing of the steps of an analysis project with an index page and links to interactive HTML reports. Create Web Report is now scriptable so the creation of the web report can be added to the end of a standard report script.
• Filter support in interactive HTML reports adds filtering through selection, but does not recalculate models, for example.

• A-optimality criterion added to Custom Design. This feature is useful when the optimal design needs to focus more heavily on certain effects rather than others, since there isn’t a natural way for D- or I-optimality to do this.
• Ability to create balanced incomplete block designs (BIBD).
• Improvements to the way priors are specified when building accelerated life test designs.
• In Compare Designs, up to 10 designs can now be compared through a JSL interface, and up to five with the UI.

• Improvements to the Fit DSD algorithm and options let the platform find weakly quadratic terms and interactions that have weak heredity.
  
• New option to create a data table from color map on correlations.
• Addition of pairwise incidence heatmap to MaxDiff designs. Improved space-filling over levels of a categorical factor.
• New option to create a run order column in Custom Design.
  

• New Functional Data Explorer platform modifies functional form of sensor/signal/streaming or batch process data. With this platform, clean up, align and conform data coming from sensors, build models using a variety of techniques and use output models  and features in other predictive modeling platforms. PRO
• Text Explorer enhancements include:
  • Support two Asian languages (Chinese and Japanese).
  • Better filtering options for tabular output, including terms, phrases and report lists.
  • Per topic word clouds.
  • Support for discriminant analysis. PRO
  • Support for validation column in the platform launch for use in a predictive modeling workflow. PRO
• Generalized Regression has new features: PRO
  • Additional distributions, including more censoring (Gamma and exponential distributions add support). 
  • Multinomial response support (nominal responses with more than three levels). 
  • Additional selection techniques, including best subset regression, backwards elimination, pruned forward selection and the Dantzig selection (which has gained popularity in the DOE community and is good for n<<p problems, liked supersaturated designs).
  • Other features, including a reset solution button, VIFs, adjusted R-square, RMSE, QQ-Plots and F-tests. Finally, custom tests are supported, cross-validation support has been added for two-stage forward selection and the platform has been adjusted to perform faster with large and/or complex problems with many effects.

• In Explore Missing Values, the addition of a new SVD/Matrix Completion-based approach incorporates validation into the dimension reduction step and scores new observations using the V matrix and a least squares estimate of the U for new rows of data. Explore Missing Values supports a validation column for imputation. PRO
• Profiler Simulator adds distribution-based simulation of response errors in Standard Least Squares and Generalized Regression.
• Distribution platform: Addition of SHASH distribution.
• Extended support for Multiple Response modeling type.
• Adjustment made to the PCA algorithm using a sparse implementation that makes computing principal components much faster (even on non-sparse data). Also support for supplementary variables.
• Fit Stepwise uses transforms instead of formula columns for saving terms for parts of categorical effects.
• K-NN and Naïve Bayes now have profilers. K-NN adds model selection to interactively change K and explore various model fits. Improved model diagnostics available in Naïve Bayes platform. PRO
• Performance improvements to Formula Depot, including ability to publish recode columns, more model support and support for derived columns. PRO

• Drift Detection, Goal Plot, three-way chart metric and numerous UI enhancements in Process Screening.
• New platform for managing spec limits.
• CUSUM Control Chart platform added to the Quality Engineering menu.
  
• Addition of Bayesian Inference for Fit Life by X.
• For parametric survival, the addition of a nested model tests in addition to graphical comparisons to match Fit Life by X.
• Improved Fit Censored JSL function to improve performance of Life Distribution with very large data sets containing Interval Censoring.
• New Probability plot of Standardized Residuals in Cumulative Damage.
• New side-by-side comparison of fixed parameter results along with unconstrained results in the density profiler.

• Improved interface for censor indicator in failure cause column for life distribution now works like censor with dropdown list.
• Recurrence analysis includes a calendar event plot, interarrival by age plot and save simulation command for the non-homogenous Poisson process models fit within the platform. Stepwise models saved as prediction columns now require only a single column.
• Repairable system simulation, cross-component event/action links can be turned off with the diagram operation option, which is useful for cleaning up a complicated diagram. Individual links can be viewed by selecting an action or event block to view a relationship when the cross-component link option is turned off. PRO
• Three-way control charts can now be launched directly within the Process Screening platform.
• Process capability: additional distributions, measures of variability and new Process Performance Plot.

• New ways to access data:
  • New JMP interface to Python allows connection to local install of Python, sending data to Python, executing Python code from a JSL script and returning data and results from Python programs.
  • New HTTPRequest() function in JSL allows access to web data using APIs:
    • Allows you to communicate with external web servers using JSL.
    • Communication occurs through APIs (Application Programming Interfaces).
    • Used with the new JSL JSON parsing functions, HTTP Request makes it much easier to get web data into JMP with JSL.
      

• Interface improvements to Scripting Index, including the option to show search results by relevance.
• Option to save encrypted scripts to data tables.
  

• New Multiple Factor Analysis platform helps with sensory analysis, identifies similar products and outlying panelists and facilitates pre-evaluation studies, later-stage consumer panels and studies that have untrained panelists.

• Hierarchical Bayes calculations in Choice have improved starting values before the MCMC sampling begins to reduce shrinkage for variance calculations and model parameters. PRO
  

Documentation

Download a PDF of the new features in JMP and JMP Pro or view our online searchable documentation.

• New Features in JMP 14 (PDF)
• JMP Online Documentation