MDT, short for MultiwayDecisionTree
Multiway is derived from the classical decision tree algorithm--C4.5, while it adopts a multiway splitting method for numerical variables.That is, when finding a best splitting variable, MDT fisrt applys a local entropy-based discretization method for each numerical variables. Regarding each variable as categorical, and the multiway splitting method used in C4.5 is appliable now.
This algorithm is still being developed.
##usage
python MDT.py -d dataset.csv -o output.txt -c 0 0 0 1 0 0 -g 0.4
------------argument list---------
-d,--dataset: the name of dataset
**-o,-output **: the name of output file
**-c **: binary array, categoical_predictors. For instance, 0 0 0 1 0 0 means the 4th variable is categorical
**-g **: min_gain, used in pruning tree