A multilevel Bayesian framework for predicting municipal waste generation rates.

Prediction of waste production is an essential part of the design and planning of waste management systems. The quality and applicability of such predictions depend heavily on model assumptions and the structure of the collected data. Ordinarily, municipal waste generation data are organized in hierarchical structures with municipal or county levels, and multilevel models can be used to generalize linear regression by directly incorporating the structure into the model. However, small amounts of data can limit the applicability of multilevel models and provide biased estimates. To cope with this problem, Bayesian estimation is often recommended as an alternative to frequentist estimation, such as least squares or maximum likelihood estimation. This paper proposes a multilevel framework under a Bayesian approach to model municipal waste generation with hierarchical data structures. Using a real-world dataset of municipal waste generation in Denmark, the predictive accuracy of multilevel models is compared to aggregated and disaggregated Bayesian models using socio-economic external variables. Results show that Bayesian multilevel models outperform the other models in prediction accuracy, based on the leave-one-out information criterion. A comparison of the Bayesian approach with its frequentist alternative shows that the Bayesian model is more conservative in coefficient estimation, with estimates shrinking to the grand mean and broader credible intervals, in contrast with narrower confidence intervals produced by the frequentist models.

Multi-site household waste generation forecasting using a deep learning approach.

Forecasting household waste generation using traditional methods is particularly challenging due to its high variability and uncertainty. Unlike studies that forecast waste generation at municipal or country levels, household data can present rapid short-term variations and highly non-linear dynamics. The aim of this paper is to investigate the advantages of using a state-of-the-art deep learning approach compared to traditional forecasting methods. We apply a multi-site Long Short-Term Memory (LSTM) Neural Network, to forecast waste generation rates from households using a long-term data base. The model is applied to historical data of weekly waste weights from households in the municipality of Herning, Denmark, in the period between 2011 and 2018. Results show that using a multi-site approach, instead of an individual fit for each household, can improve forecasting performance of the LSTM model by 28% on average, and that the LSTM approaches can effectively improve the results by 85% on average compared with traditional methods such as ARIMA.