This study introduces an advanced statistical methodology for analyzing Delphi outputs in scenario development, employing a novel combination of network-based fuzzy clustering and centrality measures. The approach captures intricate relationships and the possibility of overlapping clusters by conceptualizing Delphi outputs as nodes within a network, with edges defined by correlations among expert responses. Unlike traditional clustering methods, this framework allows items to belong to multiple clusters with varying intensities, mirroring the complex interconnections characteristic of future scenarios. Additionally, integrating community detection algorithms, such as Louvain’s method, and centrality metrics enhances the analysis by uncovering the network’s structural properties, offering deeper insights into constructing scenario narratives. The proposed methodology is demonstrated in a case study on future family dynamics in northeast Italy, where expert assessments of social and demographic trends inform scenario development. By addressing the key limitations of conventional Delphi clustering approaches, this network-based framework improves both the robustness and interpretability of scenario planning, serving as a versatile tool for decision-makers grappling with complex and uncertain futures.
A network-based fuzzy clustering approach to classify Delphi outputs in scenario development
Calleo Y.
;Di Zio S.
2025-01-01
Abstract
This study introduces an advanced statistical methodology for analyzing Delphi outputs in scenario development, employing a novel combination of network-based fuzzy clustering and centrality measures. The approach captures intricate relationships and the possibility of overlapping clusters by conceptualizing Delphi outputs as nodes within a network, with edges defined by correlations among expert responses. Unlike traditional clustering methods, this framework allows items to belong to multiple clusters with varying intensities, mirroring the complex interconnections characteristic of future scenarios. Additionally, integrating community detection algorithms, such as Louvain’s method, and centrality metrics enhances the analysis by uncovering the network’s structural properties, offering deeper insights into constructing scenario narratives. The proposed methodology is demonstrated in a case study on future family dynamics in northeast Italy, where expert assessments of social and demographic trends inform scenario development. By addressing the key limitations of conventional Delphi clustering approaches, this network-based framework improves both the robustness and interpretability of scenario planning, serving as a versatile tool for decision-makers grappling with complex and uncertain futures.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
