Modeling a waste disposal process via a discrete mixture of count data models

We propose a modeling framework for the data generating process of waste disposal in recyclable waste containers. It is based on a discrete mixture of count data models representing populations depositing di erent quantities in the containers, thus reflecting a realistic underlying behavior. It is tested on real data coming from ultrasound sensors mounted inside the containers and exhibits better in- and out-of-sample performance compared to a simple count data model assuming only one deposit quantity. The purpose of the mixture model is to forecast container waste levels at a future date when collection will take place. It thus becomes the first-step ingredient in a framework for e cient waste collection optimization.