INTERNAL USE ONLY: This document is internal use only at this time.
New features in JMP® 13 and JMP® Pro 13
If you like digging around in your data to find patterns and trends, you’ll love doing so with the newest release of SAS’ statistical discovery software: JMP® 13 and JMP® Pro 13. Now you can play with unstructured data as well as structured, use virtual joins to make a light-weight connection between tables, build dashboards in a few clicks, and a whole lot more. You could say that this is your lucky day. The PRO icon indicates additional features exclusive to JMP Pro 13.
Data Import, Data Table, Cleanup, Data Visualization, General
- JMP Query Builder: Ability to perform multi-table queries and joins of JMP data tables. This helps you discover interesting data and make correct queries that are repeatable and sharable. Additionally, you can bring in data from various external sources (e.g., Excel, JMP, Oracle, SQL data) and join them in memory using JMP Query Builder. Also included in the Query Builder is filter enhancements for categorical variables, expanded preference settings for categorical filters, first N rows sampling, run on open option, auto-join improvements and more. (BLOG POST, Eric Hill)
- Improved support for "Big Data" databases including: Apache Hive, Cloudera Impala, Amazon Redshift, Amazon Aurora and Amazon MariaDB.
- Modeling Types have been expanded with None, Multiple Response and Unstructured Text. (BLOG POST, John Sall)
- HDF5 file import: handles simple types of HDF5 files with dimension 2 (data tables without vector cells).
- File > Quick Open lets you search for JMP files by name without having to navigate through the file browser (Windows only).
- Support for Virtual Joins: Join two or more JMP tables without having to make an actual join through the Link ID and Link Reference column properties. (BLOG POST, Xan Gregg or Chung-Wei Ng)
- New HTML-based help system with improvements to search, indexing and speed. (BLOG POST)
- Play button to run scripts in a data table.
- Projects: A new projects system is available, accessed from File > New > Project. Projects are a way to organize, within JMP, a collection of objects such as data tables, reports, scripts, journals and logs. They may also contain external files, such as .txt, .pptx, and image files. Useful for anyone who likes to use JMP with maximized windows or suffers from window management pains. Projects have a tabbed interface, an integrated log, and can be archived and shared with others. (BLOG POST, Xan Gregg) Defer to 14.
- Retooled formula editor makes it easier to create and view formulas. (BLOG POST)
- Modeling utilities have been moved to the Screening submenu of the Analyze menu. Predictor Screening (formerly known as Screen Predictors) joins the Screening submenu along with Response Screening and the Process Screening platforms. (BLOG POST after launch only)
- JSON data can be imported and exported.
- The Excel Import Wizard includes options for importing cell coloring, specifying separators in multiple-row headers, stacking data and duplicating headers in spanned rows.
- Parallel plot element available in Graph Builder. (BLOG POST together with next bullet, Xan Gregg)
- Improvements to the Treemap element in Graph Builder: support for n-levels, drill down by clicking a subcategory and options for single color tree maps. Also, label and font improvements for creating easy-to-read, complex treemaps.
- General improvements in Graph Builder: Allow specification of alpha level for confidence intervals. Add right-click menu item for Order By. Add option to position legend inside the graph (left or right). Line element has Connect Through Missing options. New Legend Settings menu item. Add Centered Step option to Line Element. Add Geometric Mean. Add Ellipse option to show correlation value. Add Response Axis property to Points, Regression, Smoother, Caption and Formula. For Line and Bar, tolerate absent X variable. Bar was treating no X as a single point at 0. Now it is carried as a single point with a missing value, which the Line and Bar now handle. Allow up to two color and size variables in Graph Builder so that different elements can use different coloring or sizing. Add Show Labels option to gradient properties.
- New transforms: Quantile, Day of Week Abbr, random sampling, moving average, lag.
- New lightness constraints in legend settings dialog.
- New color themes: Viridis and Magma.
- New custom number format option, especially useful for axis labels.
Design of Experiments
- Fit Definitive Screening available in DOE menu. Specifically tailored modeling technique for Definitive Screening Designs (DSDs). (BLOG POST, Brad Jones)
- The DOE menu has been modified with many designers grouped under new, descriptive headings. (include in BLOG POST about menu changes)
- Addition of Number of Extra Runs options to the DSD. This creates a design that has all the alias properties of a DSD, but analysis is even more powerful with the extra runs. Option for 0, 4 or 8 extra runs, which is equivalent to adding extra factors and dropping them. (part of DSD BLOG POST)
- Compare Designs platform: In previous versions of JMP, it was difficult to compare designs easily -- especially with more than two designs. Now, a single report lets you make side-by-side comparisons across multiple designs. (BLOG POST, Ryan Lekivetz)
- Custom Design can now simulate realistic data for experiments, generating count- or pass-fail-based results with the addition of Binomial and Poisson distributions to the Simulate Responses tool. DOE simulate script is also saved to the design table so that effects and response distributions can be adjust after the design has been created. (BLOG POST about simulation along with PRO bullet below, Ryan Parker)
- Addition of Max Diff (best/worst) design creation in the Consumer Studies submenu of the DOE menu.
- Simulate response outputs formulas for use with general simulation functionality. (PRO) (BLOG POST)
Statistics, Predictive Modeling and Data Mining
- Text Explorer: New platform to analyze unstructured text data. It provides basic tools for analyzing free text, engineering notes or surveys. Regular Expressions (RegEx) interface for creating and storing custom RegExs for extracting part numbers, words between HTML tags, phone numbers, times, currency and more. (BLOG POST, John Sall) -- Updated 4/27/2016
- Modifications to Fit Least Squares default layout: Simplification of report output, Studentized Residual Outlier plot and Box-Cox transformations refit/replace options. (BLOG POST)
- Fit Model and Distribution now support multiple modeling types. (add to modeling types blog post, see above)
- Latent Class Analysis (LCA) platform added. LCA is a clustering technique for categorical variables. (BLOG POST about clustering submenu, Chris Gotwalt)
- Much improved SVD imputation (through a sparse implementation) in Explore Missing Values, which can be much faster than previous versions of this algorithm if your data contains lots of zeros.
- Addition of TOST equivalence tests in the Distribution platform.
- New curves added to the Fit Curve platform: Probit curve (2P and 4P versions); Rodbard and Hill curves. Fit Curve has its own launch now instead of requiring the Nonlinear platform to launch Fit Curve. (BLOG POST)
- Sparse PCA method addition to PCA.
- ARIMA and SARMIA now explicitly state the mean of the series in the Time Series platform.
- Time series; "interactive forecasting" added. Transfer function allows more convenient method to input values. Forecasting panel is upgraded to provide a feature to interactively drag data points in graph to specify values. Data import option also added.
- Addition of a Profiler to the Partition platform.
- New Clustering submenu in the Analyze menu includes hierarchical, k-means, normal mixtures, LCA and cluster variables. (cluster variables PRO)
- Fit Model logistic regression with nominal or ordinal variables includes an Indicator Parameterization Estimates options. This option gives parameter estimates for the model where nominal columns are coded using indicator parameterization (the SAS GLM parameterization) and are treated as continuous.
- Text Explorer analytics: Use multivariate and dimension-reducing techniques to transform, often for use in predictive modeling, the results obtained in the feature exploration phase. (PRO) (BLOG POST, John Sall) 4-27-2016
- Cox mixture models now work when there are process variables and when there is pseudocomponent coding. These models are not supported, however, for Scheffe cubic models or when there are terms higher than order 2. Addition of Cox component effects report. Hidden column for the gradient at the reference mixture so that a partial or constrained effect can be created.
- New modeling launch options in Boosted Trees and Bootstrap Forest, notably the ability to set the random seed and suppress multithreading for reproducible model fits. Also the ability to use a tuning design table and the addition of stochastic gradient descent. (PRO)
- Addition of Formula Depot platform to store and manage formulas. Generates score code for SAS, SQL, Python, Javascript and C. (PRO) (BLOG POST)
- Publish commands parallel to save commands for formula column scripts in many fitting platforms. (PRO)
- Double Lasso option added to the Generalized Regression personality of Fit Model: Performs an initial Adaptive Lasso pass for a screening and then uses the results' parameters to form weights for a second Adaptive Lasso pass. (PRO) BLOG POST with all Gen Reg features, Clay Barker)
- Addition of Two-Stage Forward Selection option to the Generalized Regression personality of Fit Model: Initial forward selection considers only main effects, then a second pass of forward selection for the interactions and higher-order terms. This technique has excellent properties for model selection of DOEs. (PRO)
- Generalized Regression general improvements: New model selection criteria, ERIC, which has been designed specifically for penalized regression problems, full model upon launch, ROC, lift curves and confusion matrices; CDF and quantile Profilers (like parametric survival), support for no intercept models and ordinal predictors, ability to save simulation formula that can be used in the Simulate utility. (PRO)
- Naive Bayes, a new method available in the predictive modeling submenu of Analyze. (PRO)
- Gaussian Process: Faster fitting of large data sets using composite likelihood methods and the Fit Mixed computational framework. (PRO) (BLOG POST)
- Fit Mixed: Interdependence, Toeplitz and Heterogeneous variance models; Indicator Parameterization Estimates option. (PRO)
- General Simulation functionality: Power calculations for almost anything (e.g., Logistic regression, Mixed Models, etc.). Support for parametric bootstrapping and randomization testing. (Add to BLOG POST on simulation, see above) (PRO)
Quality Engineering, Reliability and Six Sigma
- Process Screening platform joins the suite of Screening tools, a submenu of the Analyze menu. Process Screening performs control chart calculations for thousands of columns and/or groups. (BLOG POST, John Sall on big data problems & screening)
- Process Capability platform supports automated distribution selection (letting the data decide the best distribution); addition of probability plots for assessing distribution model fits.
- Support for non-normal capability analysis.
- Concurrent and Parallel systems reliability growth tabs added to Reliability Growth platform.
- Location-scale distributions added to Fit Parametric Survival, and a Fit All distribution.
- Addition of Cumulative Damage platform for analyzing time-varying-stress accelerated life test models.
- Fit Parametric Survival Fit All Distribution with Cause models will produce reports that are hierarchically organized.
- Life Distribution supports censored observation plot for right censored data. Turned on by default.
- Likelihood interval can be set as the default through a preference in Fit Life by X.
- Time to Event Format launch of Reliability Forecast allows user to arrange arbitrary existing risk sets.
- Recurrence Platform additions - TBD. Currently on target for JMP 13
- Addition of the Repairable Systems Simulation platform (RSS). (PRO)
- Parametric Survival supports variable selection through a bridge to the Generalized Regression personality of Fit Model using a launch button in the report. (PRO)
- The Generalized Regression personality of Fit Model now handles censored data allowing you to do variable selection with survival/reliability data; support for Cox Proportional Hazards in Distribution drop-down menu. Supports common reliability distributions including ZI. (PRO)
- Addition of Reliability Overlay by Component graph to the Reliability Block Diagram (RBD). (PRO)
- Legend in Reliability Block Diagram (RBD) can be customized for color and line type. (PRO)
Consumer and Market Research
- Choice Models: A "No Choice" option for the situation where the study participant does not want any of the proposed choices.
- MaxDiff (best/worst): A new platform providing a variation of choice analyses for ranking customer preference.
- Addition of the MaxDiff design under DOE > Consumer Studies > MaxDiff Design.
- Rate Per Response and Multiple Responses in outer nests added to the Categorical platform.
- Multidimensional Scaling (MDS) platform addition to the Consumer Research submenu of the Analyze menu.
- Association Analysis (Market Basket), a new method available in the Screening submenu of Analyze. (PRO) (BLOG POST about evolution of platform, Melinda Theilbar)
- Hierarchical Bayes analysis in Choice. (PRO) (BLOG POST)
Building Dashboards, Sharing and JSL
- The Dashboard Building mode of Application Builder is now the default. File > New Dashboard will bring up a template screen for easy building of a custom dashboard of JMP reports and data tables. Dashboards provide an interactive way to combine reports into a single window and can become a visual tool that lets you run and present reports on a regular basis with no scripting required. (BLOG POST, Dan Schikore)
- Combine Windows now includes options for viewing a combined report in summary view and/or adding a graph (selection) filter on creation. Combine Windows now creates a JMP dashboard in a single click.
- View > Create Web Report option creates a web page in which reports, descriptive text and graphics are displayed. The generated folder, index page and resulting interactive HTML reports can be uploaded to a web server for sharing interactive JMP reports grouped by date or project, for example. (BLOG POST)
- Partial support for interactive HTML reports of Graph Builder "done" state. (include in Graph Builder improvements BLOG POST)
- Interactive HTML support for dashboard layouts built in Dashboard Builder.
- Value label support for interactive HTML reports.
- Interactive HTML pre-flight report for unsupported JMP elements generated to the JMP log.
- Open data tables, journals and scripts are autosaved at regular intervals when the Autosave Timeout preference is set on the General page.
- The Create Excel Workbook in the View menu enables you to save multiple JMP data tables in a single Microsoft Excel workbook. (BLOG POST: post-launch blog posts done by Brian Corcoran already)
- Support for classes in JSL. Supports functions, methods (virtual), inheritance (this, super), messages, namespace code blocks. (BLOG POST: post-launch)
- Document-specific style support in script windows.
- Addition of the Using Namespace (); function.
- Significant optimization of the ways JSL handles lists.
- Fixed an inconsistency in handling matrix subscripting when the subscript itself was also a matrix. Using an N-item matrix subscript returns an N-item matrix. Before, a number was returned for N==1.
Callout boxes --
Pg. 1 Online:
Details on the new features
Get the full documentation of all of the news features in JMP 13 and JMP Pro 13
jmp.com/release-notes
Pg. 2 Online:
Join the JMP® User Community
Ask a question and get answers from JMP experts around the world. Share and download add-ins, scripts and sample data. Explore the Academic Community and special areas for Discovery Summit resources, JMP administrators and regional users groups.
community.jmp.com
Pg. 3 Online:
Watch Mastering JMP® webcasts
Expand your skills through our complimentary webcast series for JMP users of all levels.
jmp.com/mastering
Pg. 4 Online:
Subscribe to the JMPer Cable®
It's the in-depth, technical newsletter for JMP users. Read it online or receive a printed copy.
jmp.com/jmpercable
