The dataset format is as follows:

Each line in the file corresponds to one instance/object. Each column in the file corepsonds to an attribute, with the final column being the decision class. For example:

  1.1  1.4  5.3  6.2  2

This is one object, with 4 conditional attributes and 1 decision
attribute. The decision class for this object is '2'. The value of the second attribute is 1.4.

For the fuzzification files, the format is as follows. Each line takes the form:

<attribute number> <fuzzy set type> <label> <values that define the set>

Depending on the fuzzy set type, there will be a different number of values expected to define the set. A description of valid fuzzy sets are below:

LSH - left shouldered fuzzy set. This requres 2 values.
RSH - right shouldered fuzzy set. This requires 2 values.
TRI - triangular fuzzy set. This requires 3 values.
TRA - trapezoidal fuzzy set. This requires 4 values.

I've attached a file to this email that will hopefully explain things better!

For instance:

  0 LSH Low 31 37
  0 TRA Medium 31 37 58 72
  0 RSH High 58 72
  1 LSH Low 31 37
  1 TRA Medium 31 37 58 72
  1 RSH High 58 72

This defines fuzzy sets for two attributes (0 and 1). Each attribute has 3 fuzzy sets defined for it: one left-shouldered, one trapezoidal and one right-shouldered.