FID is a program which generates a fuzzy logic-based decision tree, from fuzzy or symbolic data. The tree can then be used to classify data, with unknown classification, using several different methods of inference. Partly supported by NSF IRI-9504334.
Developed by Cezary Z. Janikow
GUI for FID3.4 GUI34 developed by L. Finney

Current version

FID 3.4 ( build 05/15/2005)

Key Features :

fuzzy-set based and exemplar-based inferences

fuzzifying unrecognizable data to force classification

choice of multiple norms, with the ability of selecting the best for given data

processing nominal, fuzzy, and ordered symbolic data

ability to deal with missing or noisy features, both in training and in classification data

two discretization methods for unpartitioned continuous domains:

top-down data-driven selective discretization

bottom-up data-driven global discretization

tree pruning

History of changes

FID 4.0

- Fuzzy Decision Forest plus built-in cross validation
- currently not available, will be redone as 4.1 in the future

FID 3.4

- redone inferences
- redone top-down partitioning with an additional parameter
- fixed bottom-up partitioning
- added pruning
- improved Users Manual
- (most) bugs fixed

FID 3.3

- debugged and modified top-down discretization
- added bottom-up discretization
- improved treatment of examples with missing features
- improved parameters

FID 2.1

- fixed treatment of examples with missing features
- renamed and redefined minInfPkN parameter

FID 2.0

- the old version working system made available in 1997

Free Web Site Counter