The identification of ‘fuzzy profiles’ through the c-means clustering

The numerous concepts of socio-economic hardship are, furthermore, attributable to a traditional distinction between absolute and relative conditions of hardship. The options of scientific research were therefore oriented towards the establishment of a multi-dimensional approach, sometimes abandoning dichotomous logic in order to arrive at fuzzy classifications in which each unit belongs and, at the same time, does not belong, to a category. A multidimensional index that considers hardship as the overall condition of being disadvantaged and deprived seems the most appropriate in view of the socio-economic differential analysis of demographic phenomena. The approach used in this work to synthesise and measure the conditions of the hardship of a population is based on a clustering procedure (fuzzy c-means) aimed at outlining various not defined a priori profiles, which should be assigned to each family with different socio-economic behaviours. In comparison with conventional methods, this clustering method allows a set of data to belong not only to a main cluster but also to two or more clusters with ‘fuzzy profiles’.