Algorithm::DecisionTree - A pure-Perl implementation for
constructing a decision tree from multidimensional training
data and for using the decision tree thus induced for
classifying data.  The decision tree is induced from the
training data supplied through a diskfile.  As the
documentation explains, the training data must be in a
format that shows the names of the classes, the names of the
features, their possible values, etc., at the head of the
training data file.  Please see training.dat in the examples
directory.

You can generate your own training data by specifying the
class names, the feature names, the names to be used for
feature values, etc. All this information is supplied to the
data generator in the form of a parameter file. Please see
param.txt in the examples directory for an example.

From the standpoint of practical usefulness, note that the
classifier carries out soft classifications.  That is, if
the class distributions are overlapping in the underlying
feature space and a test sample falls in the overlap region,
the classifier will generate all applicable class labels for
the test data sample, along with the probability of each
class label.

For installation, do the usual

    perl Makefile.PL
    make
    make test
    make install

if you have root access.  If not, 

    perl Makefile.PL prefix=/some/other/directory/
    make
    make test
    make install

Contact:

Avinash Kak  

email: kak@purdue.edu

Please place the string "DecisionTree" in the subject line
if you wish to write to the author.  Any feedback regarding
this module would be highly appreciated.