A Comparison of Machine Learning Algorithms in Predicting Lithofacies: Case Studies from Norway and Kazakhstan

Defining distinctive areas of the physical properties rocks plays an important role in reservoir evaluation and hydrocarbon production as core data are challenging to obtain from all wells. In this work, we study lithofacies values using machine learning algorithms determination classification various well log Kazakhstan Norway. We also use wavelet-transformed identify geological data. Numerical results presented for multiple oil gas which contain more than 90 released wells Norway 10 field. have compared including KNN, Decision Tree, Random Forest, XGBoost, LightGBM. The model score is conducted by metrics such accuracy, Hamming loss, penalty matrix. addition, influence dataset features on prediction investigated algorithms. result research shows that Forest has best among considered consistent with outcome SHapley Additive exPlanations (SHAP) framework.