Bayesian Networks Model Averaging for Bes Indicators

摘要: The measure of the equitable and sustainable well-being (Bes) is growing interest in last years. National Institute Statistics (Istat) provides, for Italy, a wide set indicators describing each domain that is, by definition, multidimensional concept. In this study, we propose use Bayesian networks to deal with basic composite Bes indicators. Its capability model very complex multivariate dependence structures useful describe relationships between belonging different domains and, being probabilistic expert system, estimated network could be also inference what-if analysis. all have been means hill climbing algorithm based on bootstrap resampling averaging order prevent bias due deviations from normality assumption.