Understanding municipal solid waste (MSW) generation is a key requirement for designing and optimizing MSW collection services. The present contribution proposes a statistical methodology to identify MSW generation patterns from MSW collection records. The methodology aims at finding statistically distinct household waste generation patterns within the days of the week and within months (seasonal variation). It is based on standard statistical methods (ANOVA complemented by non-parametric tests and cluster analysis). The methodology was applied to a Portuguese neighbourhood to assist in the definition of a waste sampling campaign to support the implementation of a pilot PAYT. The results showed the existence of groups with statistically distinct MSW generation patterns both at the weekly and monthly time scales. Three clusters of days of the week, with high, medium and low generation, and two clusters of months, with high and low generation, were identified. These results allowed to design and implement a customized field waste sampling campaign to estimate the MSW generated at the study site with minimal field work. Instead of implementing a homogeneous sampling campaign (equal number of samples for every day of the week and for every month), the samples were collected from the days and months that showed statistically distinct MSW generation pattern. The systematic procedure can be easily adapted to any given location, thus being a useful tool that combines statistical analysis with field collected data.
Keywords: Data collection; Experimental design; Pattern identification; Sample distribution; Seasonality; Waste forecasting.