2018 08 21: Previously, when using the exhaustive search algorithm it was not possible to set the maximum size of the number of attributes to be found in the best predictive model. Now it is possible to set that number, up to 5. This means exhaustive searches of this kind will be quicker than the previous unconstrained searches. My experience to date suggests that models with many attributes e.g more than 5, typically have a fairly narrow coverage i.e. apply only to a small proportion of all cases where the outcome is present.
I have also now indented three of the four search algorithm buttons, to emphasise that they are specialized options of the first un-indented search.
2018 03 11: Bugs have appeared when I have worked with a data set that has 597 cases and 35 attributes. Sometimes the transition from Select Data to Design and Evaluate fails and an error message shows up. Also, the generation a Decision Tree can take up to 5 minutes. When the transition fails, I have got around this by clearing out the data set in Input Data, saving the file, closing down other programs, then re-entering the data.
2018 03 09: When a Decision Tree is generated it is now possible to see a string of summary measures describing the overall performance of the Decision Tree, just above the tree itself. The performance measure on the left is the one chosen when the Decision Tree algorithm is chosen
2018 02 16 Be careful when you enter data from a file where the cell values may have hidden spaces before or after them. This will stop the Design and Evaluate worksheet from working. Do a search a and replace for blank spaces, before uploading the data set. PS: This should no longer be necessary, the code has been changed to deal with these traps.
2017 10 30 EvalC3 now works on Macs and PCs. But having no direct experience of using it with Macs, there may be bugs yet to be discovered when using Macs