A Knowledge Discovery from Data Process to Assess the Reliability of the Human Development Index
Résumé
The United Nations Development Program (UNDP) defines a set of indicators about the level of human development in countries, collects data and issues reports annually. This paper targets assessing the ability of the the UNDP Human Development Index (HDI) to reflect a factual image of human development in a country. Assessment is based on a reproducible process of knowledge discovery from data (KDD). Main results include: (i) Disparities among the sample dataset countries are most visible for the distribution of the gross national income per capita (GNI-pC) dimension; (ii) The formula used to calculate the HDI weakens the contribution of the GNI-pC, so some countries are catching up the same HDI of other countries having higher GNI-pC only due to a higher expected indicator (like expected years of schooling). This is not in favor of the HDI which is supposed to reflect factual development level while an expected measure can not reflect facts; and (iii) The clustering found in the data mining step appears factually realistic relatively to the HDI data including the high disparities in GNI_pC.
Origine : Fichiers produits par l'(les) auteur(s)