Methodical Approaches To Classification Of Clasters Of Economic Entities

Figure 1: The algorithm: (1) define a similarity matrix Mij that captures the relatedness between any two industries; (2) make broad parameter choices β; (3) use a clustering function to create a cluster grouping C based on the similarity matrix and parameter choices (C=F(Mij, β)); (4) calculate benchmarking scores for each C and identify the "best" candidate sets of cluster efinitions (C*s); and (5) assess the individual clusters in each C*, identify outlier industries, definitions C**.

< Back to article