Instructions for use:
(note that in rough set terminology, a reduct is a subset of features)

typing

        java rsar <dataset>

will generate output of the form:

        <== RSAR with bailing ==>
        X attributes in dataset.
        Y objects in dataset.

        Choosing {3} ==> 0.25
        Choosing {1, 3} ==> 1.0
        QuickReduct is {1, 3, 4}
        Time taken = 1.5E-4 mins

The line starting "QuickReduct" indicates what features should be kept
in the dataset - the rest can be removed. In the above example,
features 1, 3 and 4 should be kept. Feature numbering starts from 0,
so the first feature in a dataset is referred to as 0. The final
number appearing in the reduct is always the decision feature and will
always appear in the final subset. In the example, the decision
feature is 4 (i.e. the fifth column of the dataset). Hope that makes
sense!

The dataset format is as follows:

Rows == objects/instances. 
Columns == features. 
The final column is the decision feature (the classification).

All dataset values for RSAR must be integers and should be separated
by spaces or commas.


For example the dataset:

0 1 2 1 0 5
1 3 4 2 1 2

contains two objects, with 5 conditional features (predictors) and one
decision feature. The classification of the first object is 5. 