Fuzzy partitions may be useful to generate a distance function applicable to real values, called FP-distance. This allows the introduction of expert knowledge in distance calculations. The program generates a (n x n) matrix of all distances between the (n) rows of the current data file. By default, the distance is normalized in each dimension between 0 and 1. FP-distances and Euclidean distances can be combined in the multi-variate case. Their aggregation is done using a Minkowski combination, with the p exponent as parameter. The matrix can be exported to be introduced in clustering algorithms, for instance in R.

A color image of the one-dimensional distorsion of the Euclidean distance by the FP-based distance is available.

The FP-based distance is presented in detail in:

Serge Guillaume, Brigitte Charnomordic, and Patrice Loisel.

Fuzzy partitions: a way to integrate expert knowledge into distance calculations. International Journal of Information Sciences, page In Press, 2012.