Impact copolymers are synthesized using ethylene and propylene monomers in various proportions. The mechanical properties of these polymers are directly influenced by the ethylene content because an increase in ethylene results in significant enhancements in the copolymer properties. To precisely quantify the ethylene content, ATR-FTIR was employed, leveraging calibration methodologies and prediction models based on machine learning (ML). Principal component regression (PCR), artificial neural networks (ANN), support vector regression (SVR), and k-nearest neighbor (kNN) were used. Green ethylene in concentrations ranging from 0.5% to 53% was used. R 2 , root mean square error of calibration (NRMSE), and root mean square error of prediction (RMSEP) were used as parameters for judging the best ML model. This study demonstrated that focusing on bands between 690 and 1325 cm -1 enables effective classification and prediction of green ethylene concentrations. Even when all measured bands within this range are reduced to a space of only two principal components, a remarkable 97% of the variance is explained. The results suggest that mid-infrared spectroscopy could be a useful tool for quantitative analysis of green ethylene when machine learning algorithms are used. Keywords: Green Ethylene, k-Nearest Neighbor, Support Vector Regression, Artificial Neural Network, Fourier Transform Infrared Spectroscopy DOI: https://doi.org/10.35741/issn.0258-2724.58.6.15