New in JMP 18 and JMP Pro 18

JMP 18 and JMP Pro 18 offer many new capabilities .  See a video of some of the key new capabilites and download the Journal to try some of the techniques yourself in JMP 18. Note, for several capabilites you will need JMP Pro 18. 

Watch the whole video below, or click on titles in the bullet list to see only the segments on the topic in the list.

See how to use:

• New capabilites for tables,  reports and contextual help
  • Data Manager Using Columns Manager – Use interactive boxes to prepare data interactively. Examine, sort, change, add, and group Data. Launch analyses.
  • Encapsulated Analyses and Precalculated Statistics and Graphs in Column Headings
  • One-Click Report Customizations -  Use automatic, editable and sharable platform presets.
  • Sample Index - Interactively find sample data and analyses, Including by Subject Area.
  • Deep Menu Search – Search for Options Including Those Included in Lower- Level Menus.
• New capabilities for process optimization, modeling and reliability
  • Precision Recall Curves – Useful when you have unbalanced data.
  • Profiler Visualizations – Interact with variables to optimize for variability using overlaid interactions and examine data points to see how well data supports model interactions.
  • Fatigue Modeling - Create, view results for, and select from 24 different models, choose from six S-N curve types together with one of four distributions to assemble a model, get detailed information on, and compare, all models.

• New for Data Import, Engineering and Production

  • Pi Server Import – Interactively import multiple raw, plot or interpolated data over specific period of time.
  • Configurable Data Connectors
  • Equivalence Tests on Matched Pairs – Interactively select options for and run Matched Pairs to get new standardized view of  Equivalence Test hypotheses, results description, statistics and Forest Plots.
  • Nonparametric Equivalence Test – Use Wilcoxon method that creates non-parametric statistics to determine the equivalence.

• New  for Spectral, Peaked and High-Dimensional Data (JMP Pro)

  • Four new Direct Functional Data Explorer (FDE) Spectral Models) – Use MCR (Multivariate Curve Resolution), unconstrained MCR, penalized SVD or non-negative penalized SVD.  (JMP Pro)
  • Functional Data Explorer (FDE) Spectral Peak Finding - Use parametric modeling capabilities to automatically detect peak features (location, height, width) contained in spectral data. (JMP Pro)
  • Multivariate Embedding Update – Use non-linear dimension reduction techniques to project high dimensional data points into lower dimensional space and examine Uniform Manifold Approximation and Projection (UMAP) plots to visualize and identify patterns in high-dimensional data visualization. (JMP Pro)

• New Python Integration

  • Python Library and Integration Enhancements – Use version of Python that ships with JMP for a consistent environment that will work immediately. Get direct memory access to Python data tables in a live environment and use Python Script Editor that uses Python conventions and an embedded log. 

   

Questions answered by @HydeMiller and @MikeD_Anderson  at the live webinar.

Q: How do I use Collapsible Annotiations?

A: From a graph in Graph Builder, start with the Annotations icon. To collapse an annotation, right click on an annotation and select Minimized. The collapsed annotation will appear as an “i” icon. See video below.

Q:  Please clarify how confidence intervals are used?

A:   The shaded zones on the Profiler are the confidence interval on the line itself. So, there's a line of fit through the data and the confidence interval shows where that line of fit it could be. That is confidence on the line. It's not a confidence interval on the values. Now, if you go to the Red Triangle menu, Prediction Intervals are available and they show the confidence interval for the predicted values themselves. There is an  aqua color line, and then the numbers that correspond to the lines that show the confidence for the prediction interval. In JMP 18 you now can kind of show both of them on the Prediction Profiler.  See video:

Patrick Guiliano @PatrickGiuliano , JMP Technical Support sent this information after the webinar:

The following commentary came from another JMP user. The "Confidence" lines (see image) are slightly curved, and define a 2-dimensional space in which we claim the true linear regression line exists; that is, if we imagine that the observed linear regression line can rotate about the point that is Xavg, Yavg but only within that space (i.e., the rotating line has to stay completely within the confidence lines),  then the true parameter slope of the line is said to be somewhere between the largest and smallest slope possible by performing such rotations, at a confidence level = to whatever confidence value we used to calculate those confidence lines. The confidence lines can therefore be used to predict the worst-case scenario for an AVG Y-value at a SPECIFIC X value.

The Prediction interval lines are also slightly curved; they define a 2-dimensional space in which we PREDICT the next single X,Y-value we obtain will occur. Our prediction is more confident, the closer we are to the Xavg,Yavg point, and is less confident the further away we are, which is why the lines looked curved.

Q: Do the equations come over to use Profiler?

A:  Not with this release. They were focusing on the core framework. You could add a request for that to the JMP wishlist for a future release.

Q:  Is it possible to call JMP from Python? IE the opposite of what you showed, where JMP calls Python.

A: There are certain calls that go either direction. The documentation will indicate  what all is allowed.

Resources

• Developer Tutorial: Analyzing Fatigue Testing Data using JMP Fatigue Modeling
• Developer Tutorial: New Python Integration and Scripting Capabilities in JMP 18