Fuzzy projection pursuits (Q912543)

Two objections to the ``classical'' approach in projection pursuit have inspired the authors to translate the classical projection pursuit technology for a fuzzy-reasoning approach. These are: (i) The classical approach takes into account all axes (all components of data) with the same weights not assuming different importance, precision etc., (ii) the successive application of LS on projected data may not be appropriate (but the present reviewer hesitates whether the LS approach is the only one which is usually employed; also some robust methods are frequently utilized). Some index of interestingness is proposed. Projections into a plane and on a straight line are considered separately. An index of interestingness is based on a ``projection of membership functions''. An example of projection of a plane on a line is given.