The fuzzy clumping method

This study proposes a new method of fuzzy clustering which generates a hierachical series of fuzzy clumpings. The units are gathered together in such a way as to obtain overlapping group. Using the fuzzy clumping method much of the information obtained may be retained which is lost during the non fuzzy clustering of the set of units. This may be traced to the fact that the fuzziness can grasp and render explicit the very imprecision which is an intrisic feature of every clustering and thanks to the overlappings, it is not necessary to impose limits on the units aggregated. For a more concrete assessmet of the descriptive capacity of the method illustrated and to show how the results obtained may be analysed, an application was performed of the procedure proposed on cause-specific mortality data in 22 European countries. The results obtained confirm the outcome of other methodologies as well as corroborate previously expressed speculations regarding mortality.