Unsupervised machine learning approach for building composite indicators with fuzzy metrics

Abstract

[EN] This study aims at developing a new methodological approach for building composite indicators, focusingon the weight schemes through an unsupervised machine learning technique. The composite indicatorproposed is based on fuzzy metrics to capture multidimensional concepts that do not have boundaries, suchas competitiveness, development, corruption or vulnerability. This methodology is designed for formativemeasurement models using a set of indicators measured on different scales (quantitative, ordinal and binary)and it is partially compensatory. Under a benchmarking approach, the single indicators are synthesized.The optimization method applied manages to remove the overlapping information provided for the singleindicators, so that the composite indicator provides a more realistic and faithful approximation to the conceptwhich would be studied. It has been quantitatively and qualitatively validated with a set of randomizeddatabases covering extreme and usual cases.