Variational Autoencoders for Assessing Sustainability

Reliable, impartial, and timely data models are urgently needed to inform global sustainability measures. As a response, implicit and explicit fuzzy methods that evaluate qualitative and quantitative variables have been proposed. Meanwhile, decision-makers are often compelled to address immediate over long-term risks. In the era of big data, artificial intelligence, and the internet we need to leverage the power of statistical models to support effective leadership. This article introduces a new technique for evaluating sustainability based on the Sustainability Assessment by Fuzzy Evaluation (SAFE) model: Variational Autoencoder plus graphical analysis (VAE&GA). This approach produces SAFE-like rankings backed by a dual axis chart that groups countries according to their most important unique indicators. VAE&GA is thus a more objective alternative to current fuzzy methods.