Cross-Validation of Circular Economy Causal Network Maps

Abstract

AbstractThe circular economy (CE) paradigm has significant potential in addressing resource depletion challenges. CE aims to attain a cyclic economy where materials are recirculated. Identification of its drivers and barriers facilitate the transition towards CE. Mapping its causal networks are essential in understanding their interrelationships. The decision-making trial and evaluation laboratory (DEMATEL) possesses the ability to map causal networks, utilizing perception on pairwise relationships for mapping. This utilization relies on subjective perceptions, resulting in biases. Fuzzy cognitive maps (FCMs) also possess the ability to map causal networks. FCMs have a similar architecture with neural networks, allowing calibration according to empirical data. Integrating both frameworks yield an empirically trained causal network. A hybrid DEMATEL-FCM framework is developed and is applied in country-level economies. The proposed hybrid approach is capable of mapping an initial causal network using DEMATEL and then training the network through FCM. The network attained an accuracy of 73.20% on training and 73.81% on testing, revealing economic attractiveness and consumer demand as significant to material footprint reduction. The DEMATEL-FCM framework is applicable to the validation of causal networks. Findings from this work may be used for the assessment of driver interrelationships for CE.KeywordsMachine learningMulti-criteria decision analysisSustainable development