The ugliest, deeply-nested part is just a machine-generated classification tree (e.g., http://en.wikipedia.org/wiki/Decision_tree_learning http://en.wikipedia.o

The ugliest, deeply-nested part is just a machine-generated classification tree (e.g., http://en.wikipedia.org/wiki/Decision_tree_learning http://en.wikipedia.org/wiki/Decision_tree_learning). Notice how all the leaf nodes (ends of if/then branches) do only one thing: set isDirty to 0 or 1. And, all the if/then conditions are just thresholds of one variable against a constant threshold. The later part seems to be exceptions to the above tree. I've never heard of this, but it seems like what they used: http://weka.sourceforge.net/doc/weka/classifiers/rules/Ridor.html http://weka.sourceforge.net/doc/weka/classifiers/rules/Ridor...