To help users interpret how predictions being explored in the profiler relate to the data used to build them, provide options to show:
- Histograms and/or box plots in the borders of the profiler. Histograms were already suggested by @FN in Add histograms to the JMP profiler . Extend the red dotted line through them.
- Time series (or row series) trends in the profiler borders with the red dotted line extended through them.
- Scatterplots in fields behind the profiler, turned on by default.
- The extrapolation statistic as an additional y/predicted variable, shown at the bottom, show the UCL on the axis
- A scatterplot matrix showing combinations of all variables - not on the drawing below
- Option to scale label sizes - useful when trying to visualize large profiler, make a profiler with 20x20 variables, shrink the cells to 1/2 the normal size, and still be able to identify variable names
With everything turned on, the user would see this along with a scatterplot matrix, again with the current values as a red dotted line through all plots (I used blue here):
@cwillden sorry it took so long to enter this after our conversation at the last Discovery!