A sustainability approach using fuzzy logic and data mining

Resumen

Sustainable development goals are now the agreed criteria to monitor states, and this work will demonstrate that numerical and graphical methods are valuable tools in assessing progress. Fuzzy Logic is a reliable procedure for transforming human qualitative knowledge into quantitative variables that can be used in the reasoning of the type “if, then” to obtain answers pertaining to sustainability assessment. Applications of machine learning techniques and artificial intelligence procedures span almost all fields of science. Here, for the first-time, unsupervised machine learning is applied to sustainability assessment, combining numerical approaches with graphical procedures to analyze global sustainability. CD HJ-Biplots to portray graphically the sustainability position of a large number of countries are a useful complement to mathematical models of sustainability. Graphical information could be useful to planners it shows directly how countries are grouped according to the most related sustainability indicators. Thus, planners can prioritize social, environmental, and economic policies and make the most effective decisions. One could graphically observe the dynamic evolution of sustainability worldwide over time with a graphical approach used to draw relevant conclusions. In an era of climate change, species extinction, poverty, and environmental migration, such observations could aid political decision-making regarding the future of our planet. A large number of countries remain in the areas of moderate or low sustainability. Fuzzy logic has proven to be an uncontested numerical method as it occurs with SAFE. An unsupervised learning method called Variational Autoencoder interplay Graphical Analysis (VEA&GA) has been proposed, to support sustainability performance with appropriate training data. The promising results show that this can be a sound alternative to assess sustainability, extrapolating its applications to other kinds of problems at different levels of analysis (continents, regions, cities, etc.) further corroborating the effectiveness of the unsupervised training methods.