Machine Learning of polymer types from the spectral signature of Raman spectroscopy microplastics data

The tools and technology that are currently used to analyze chemical compound structures identify polymer types in microplastics not well-calibrated for environmentally weathered microplastics. Microplastics have been degraded by environmental weathering factors can offer less analytic certainty than samples of exposed processes. Machine learning techniques allow us better calibrate the research analysis. In this paper, we investigate whether Raman shift values distinct enough such well studied machine (ML) algorithms learn using a relatively small amount labeled input data when impacted degradation. Several ML models were trained on well-known repository, Spectral Libraries Plastic Particles (SLOPP), contain intensity results range plastic particles, then tested aged particles (SloPP-E) consisting 22 types. After extensive preprocessing augmentation, random forest model was SloPP-E dataset resulting an improvement classification accuracy 93.81% from 89%.