Forecasting household packaging waste generation : a case study

Abstract

Nowadays, house packaging waste (HPW) materials acquired a great deal of importance, due to environmental and economic reasons, and therefore waste collection companies place thousands of collection points (ecopontos) for people to deposit their HPW. In order to optimize HPW collection process, accurate forecasts of the waste generation rates are needed. Our objective is to develop forecasting models to predict the number of collections per year required for each ecoponto by evaluating the relevance of ten proposed explanatory factors for HPW generation. We developed models based on two approaches: multiple linear regression and artificial neural networks (ANN).The results obtained show that the best ANN model, which achieved an R 2 of 0.672 and MAD of 9.1, slightly outperforms the best regression model (R 2 of 0.636, MAD of 10.44). The most important factors to estimate HPW generation rates are related to ecoponto characteristics and to the population and economic activities around each ecoponto location.