Fuzzy Classification for Farm Household Characterization

Abstract

Most household classifications use hard classification procedures that limit a household to only one cluster. In this paper, fuzzy classification, in which individuals can belong totally, partially or not at all to a particular cluster, with membership showing how well they fit in each cluster, was tested as an alternative clustering procedure. The results show that membership values, which are an extra output of the fuzzy classification, are a useful indicator of how well a particular household fits in a given cluster. Such information is useful when choosing households to use for agricultural technology testing.