This work aims to propose a general data-driven plastic waste categorisation procedure that defines their recyclability based on classification into material recycling classes. The contamination in plastics, such as metal fillers or additives, is accumulated during the entire Life Cycle, which can be harmful to either mechanical or chemical recycling. The plastic polymers can also degrade during recycling due to weakened chemical bonds in the polymers. The diversity of plastic material types and products makes it necessary to use a data-driven quality-based definition of plastic waste properties to facilitate proper waste recycling and mitigation. This study demonstrates the use of Machine Learning tools that enable automated classification to analyse the plastic waste data and derive the indicators for plastic waste recyclability. Tree-based models such as the Decision Tree Model and Random Forest Algorithm are used as they produce interpretable if-then rules for plastic waste categorisation. The proposed method allows an analysis of the metal contamination and degradation data in a collection of plastic material samples or batches to derive a general categorisation rule for a polymer type – PE. The data-driven plastic categorisation could help in understanding the current waste practices and determining a proper recycling plan for local or even global plastic waste.
