Time series forecasting: a study on local urban waste management in a Portuguese city

Abstract

Due to economic and social development in general and to population growth, the volume of waste, particularly municipal waste, has been significantly increasing in recent years. VITRUS AMBIENTE, EM, S.A. is a municipal company that operates at various levels in local business management, namely in Urban Waste Management (UWM), and ensures the collection of waste in the Municipality of Guimarces (located in the northwest of Portugal). A pioneer project termed "PayAs-You-Throw" (PAYT) was implemented in the Historic Center (HC) of this municipality and is managed by the VITRUS AMBIENTE. VITRUS is the managing entity and the Urban Hygiene Service is responsible for implementing the necessary measures to ensure the success of this project. This work focuses specifically on Urban Waste Management and aims at modeling and forecasting the behavior of urban waste production within the company's PAYT area. Time series forecasting models - BoxJenkins SARIMA models - are applied both for the estimation and forecasting of short-term organic and recyclable intramural waste collection in the Historic Centre of Guimarces in the PAYT system implementation pilot zone. The dataset consists of weekly measurements from 2016 to 2019. The methodologies used support the company's management and decision-making process regarding Urban Waste Management, aiming at improving the services provided to the population and having always as its cornerstone the preservation of the environment.