Advancing Sustainability: Machine Learning Projections of Palm Oil Product Emissions in Indonesia

Along with Indonesia's rapid development of the palm oil oil industry, various environmental and social problems have emerged. The palm oil industry is a significant contributor to greenhouse gas emissions, releasing carbon dioxide (CO
2
) and methane (CH4) during land clearing, burning of forests, and peatland degradation. These emissions contribute to global warming and climate change. The prediction of palm oil product emissions holds significant importance for several reasons, each contributing to the broader goals of environmental sustainability, climate change mitigation, and responsible industry practices. Dataset of palm oil product emissions must be analyzed to address this problem. The study aims to identify the factors that contribute the palm oil product emission and to predict the emissions of Palm oil products in Indonesia using some regression algorithms. The number of data is 200 from the Sustainable Production Systems and Life Assessment research center of the National Innovation Research Agency (BRIN). This research using correlation test for obtaining the best contribution factors to emission variables. To predict the emission is by using three algorithm models i.e., Gradient Boosting Regressor, Random Forest Regressor, and AdaBoost Regressor. The result performed that the most influential variable to the emission is Water, Bunch Ash, Paraquat, Genset, and Electricity. Furthermore, the Gradient Boosting regressor is the best algorithm to predict emissions of palm oil products with the lowest Mean Absolute Error (MAE) evaluation value of about 0.0041, Mean Squared Error (MSE) around 9.73e-05, Mean Square Logarithmic Error (MSLE) around 3.09e-05, and has the R-squared (R2) value that is closest to 1, which is about 0.9967.