Bayesian network modelling of hierarchical composite indicators

Abstract

The water, sanitation and hygiene (WaSH) sector has witnessed the development of multiple tools for multidimensional monitoring. Hierarchical and composite indicators (CI)-based conceptual frameworks provide one illustrative example. However, this approach does not address the existing interrelationship of the indicators they integrate. Bayesian Networks (BNs) are increasingly exploited to assess WaSH issues and to support planning and decision-making processes. This research aims to evaluate the validity, reliability and feasibility of BNs to replicate an existing CI-based conceptual framework. We adopt a data-driven approach and we propose a semi-automatic methodology. One regional monitoring initiative is selected as a pilot study: the Rural Water Supply and Sanitation Information System (SIASAR). Data from two different countries are processed and analysed to calibrate and validate the model and the method. Major findings show i) an improvement of model inference capacity when providing structure to the networks (according to the CI-based framework), ii) a reduction and quantification of the key components that explain a pre-defined objective variable (implying important advantages in data updating), and iii) an identification of interlinkages among these components (which might enhance multi-and trans-disciplinary actions). We conclude that BNs accurately replicates the CI-based conceptual framework. The proposal contributes to its wider application.

Data-driven Bayesian network modelling to explore the relationships between SDG 6 and the 2030 Agenda

“…In doing so, we contribute to the existing applications of BN modelling to the SDG 6. To date, BNs flexibility has been exploited to address the inherent complexity of individual SDG 6 targets, and the literature shows several studies on target 6.1 (Fisher et al, 2015;Cronk and Bartram, 2017;, 6.1 and 6.2 (Dondeynaz et al, 2013;Giné-Garriga et al, 2018;Requejo-Castro et al, 2019) or 6.4 and 6.5 (Mohajerani et al, 2017;Molina et al, 2009;. To the best of our knowledge, however, BNs have not been applied to explore the relationships between targets of different goals.…”

“…Briefly, the constraint-based method learns the network structure by analysing the probabilistic relations with conditional independence (C-Ind) tests; while the score-based method assigns a score to each candidate BN and then tries to maximize it with a heuristic search algorithm (Scutari, 2010). In this study, we have opted for the constraint-based SLA since it reduces the subjectivity of the links among the nodes of the network (Requejo-Castro et al, 2019). In addition, despite the large amount of data needed to guarantee the reliability of the C-Ind tests, a recent study points out that constraint-based algorithms are more accurate than score-based algorithms for small sample sizes (Scutari et al, 2018).…”

“…BNs flexibility is commonly exploited in scenario simulation and analysis (e.g. Bromley et al, 2005;Dang et al, 2019;Eldridge et al, 2019;Gonzalez-Redin et al, 2019;Molina et al, 2009), but has also been used to identify the key factors that influence aspects of interest (Li et al, 2019;Song et al, 2018), to replicate hierarchical composite indicators (Requejo-Castro et al, 2019), or to elucidate the network structure underlying the data-athand, through associated structure learning algorithms (SLA) (Alameddine et al, 2011;Garcia-Prats et al, 2018).…”

Assessing Landscape Fragmentation: A Composite Indicator

“…In economics, CI have been used for assessing the national economic performance of Switzerland [18], equitable and sustainable well-being in Italy [19], travel and tourism competitiveness worldwide [20], tourist country-brand ranking [1], human development in Europe through a revised weighted release of the well-known Human Development Index (HDI) [3,21], tourism destination competitiveness worldwide [22], and economic development in Iran [23]. In engineering sciences and regional planning, CI have been adopted for studying security of energy supply in the European Union [24], the effectiveness of transit options in Vancouver, Canada [25], the performance of sanitation and hygiene systems in Nicaragua and Honduras [26], the functionality of hydropower plants in Brazil [27], the level of rurality of municipal Italy [28], and the performances of regional planning policies of a sample of Italian municipalities [4]. Sustainability is an overarching concept encompassing three pillars (i.e., environment, economy, and society) and a variety of related disciplines.…”

Climate adaptation for rural water and sanitation systems in the Solomon Islands: A community scale systems model for decision support

“…A BBN is a type of system model that is particularly suited to using sparse data and handling uncertainty to address the issues of multiple complex hazards, and is increasingly being used in the context of WaSH (e.g. Dondeynaz et al 2013, Phan et al 2016, Giné-Garriga et al 2018, Requejo-Castro et al 2019. In this instance it is used to support decision making to increase the availability of safe drinking water in rural communities in the Solomon Islands.…”