We are happy to announce the latest release of DataModeler Pro. The complete release notes are below; however, the most obvious new addition is support for automatic pagination of long tables by GridTable. This ripples automatically into many display functions — for example, the DataSummaryTable below — so that tables will fit on a screen to facilitate interactive exploration.
If you would rather see all of the table content, you can, of course, have the previous behavior; however, if you set PageView -> Print, then the table will be split into the specified MaxRowsPerPage for easy export into other documents. The pagination defaults for many functions have been set to be intelligent choices. Along the way, we also fixed the ToolTipFunction for ModelSelectionReport and ModelSelectionTable so that, for example, a ModelPredictionPlot can be used as a popup.
We have also enhanced CreateLinearModel to have parity with SymbolicRegression in terms of being able to handle non-numeric data sets and support for AllowedVariables, ExcludedVariables and MetaVariables. Any developed models can be processed by ConvertToFittedModel to allow conventional analysis tools like ANOVA tables to be easily generated. (Mathematica’s models embed the entire data set in them so you don’t want to convert every DataModeler model if you have big data sets or lots of models since there is a memory footprint impact.)
Speaking of memory footprints, we worked around a design feature in Mathematica’s Fit function since there appeared to be a leak if trig or exponential functions were being used in model development (which is not the default) and the OptimizeLinearModel option was enabled (which is the default behavior) during SymbolicRegression.