A hierarchical systems modelling approach based on neural networks for forecasting global waste generation: a case study of Chile

Abstract

In this-first every study for Chile, a neural network based hierarchical modelling approach is proposed for forecasting domestic waste generation for the whole country. Over 30 global variables from the 342 communes (municipalities) in the country were analysed extensively using statistical tools that led to 5 significant explanatory variables: population, percentage of urban population, years of education, number of libraries and number of indigents. The five explanatory variables were used to develop a feedforward neural network for predicting volume of global waste generation for a particular year (2002 in this case) in Chile and assessing the contribution of variables. The model had validation R² of 0.82.