Before building a regression model (linear regression, regression tree, or others), you may want to create discretized versions of some of your continuous predictors. This add-in provides tools for doing supervised binning (using the response column) and unsupervised binning (ignoring the response column). After specifying the maximum number of bins that you would like, the optimal number of bins is chosen using the Bayesian information criterion and binned versions of the specified predictors are added to the original data table. This add-in was designed using new features in JMP 10 and will not work with previous versions of JMP. For example, open the Boston Housing data in the sample data folder and try binning crim, nox, and age using mvalue as the response.
For more information about this add-in, see this blog post.
