Subscribe Bookmark RSS Feed



May 28, 2014

Naive Bayes Add-In

An add-in for computing Naive Bayes probabilities, developed by Ian Cox (Ian@JMP) and BradyBrady for academic purposes.

1.  Open a data table with a categorical outcome and categorical predictors (Note:  All variables must be coded as Character and Nominal).


2.  Specify the variables, and click OK


     JMP Produces:

  • A data table, with calculations to compute the predicted probabilities and the most likely outcome.  Note that the most likely outcome is based on a cutoff of 0.50.
  • A cell plot, which provides a graphical representation of the results
  • An interactive dialog for computing predicted probabilities for particular features or conditions

3.  Select the features of interest, and click Compute Probabilities.


Updated 10/2/2015:  Will now accept numeric columns, but these columns must be coded as Nominal.  The Value Labels column property must be removed prior to using the add-in.  Continuous predictors must binned or recoded (and require the nominal modeling type).


This is cool! Have you considered providing a Tree Augmented Naive Bayes (TAN), which outperforms naive Bayes, yet at the same time maintains the computational simplicity (no search involved) and robustness that are characteristic of naive Bayes?


Thanks!  We haven't discussed adding TAN, but thanks for bringing this to our attention!  FYI - Naive Bayes is a new feature in JMP 13, which will be released in September.




I love the Naive Bayes Add-In!  It works great at my office with JMP 11.0.0 under Windows 7 Enterprise.  However, I get the following error at home with JMP 11.0.0 under Windows 8.1:


Subscript Range in access or evaluation of 'Subscript' , scoreColList[i]


Does anyone know what's going on?





California State University, Channel Islands

Article Tags