Bayesian Optimized Machine Learning Models for Predicting Gas Flaring Volumes in Nigeria's Oil and Gas Sector

This study aims to understand the application of Machine Learning algorithms such as Support Vector Regression (SVR), Random Forest, and K-Nearest Neighbors (KNN) in predicting gas flaring activities in Nigeria’s oil and gas sector. This research examined models’ performance with both default and Bayesian Optimized hyperparameters, with the best Bayesian optimized hyperparameters set through training. According to the results of experiments, all models have performed remarkably well; however, the SVR model outperformed the others with an R-squared score of 0.993, MSE of 0.657. The optimized versions of the Random Forest and KNN models also performed well. This serves as a testament to the growing capabilities of technology through Machine Learning to estimate nonlinear flare co2 volume functions, which improves environmental, regulatory, and legislative monitoring in Nigeria’s petroleum industry.