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

Abstract

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

Soil Salinity Classification Using Machine Learning Algorithms and Radar Data in the Case from the South of Kazakhstan

“…The result of SHAP is presented for the Random Forest model since SHAP does not have the ability to use the Gaussian Process model. The use of the library has a wide range of applied tasks (Merembayev et al, 2021;Amirgaliyev et al, 2019;Muhamedyev et al, 2020) and is an effective tool for understanding a model. SHAP is a kind tool for explaining the various patterns, and it gives a significant value to each feature.…”

Occupational Injury Risk Mitigation: Machine Learning Approach and Feature Optimization for Smart Workplace Surveillance

“…The method will determine the similarity between the new data and available data and group the new data into the most similar categories to the existing data. The algorithm work by [ 32 ], (i) selecting the number of K of the neighbors, (ii) computing the ‘Euclidean distance’, which is to measure the distance between any two points. The formula as in Equation (2), and (iii) from the calculation in (ii), the category for a new data point is assigned to the maximum number of neighbors.…”

Gamma-ray log derivative and volatility attributes assist facies characterization in clastic sedimentary sequences for formulaic and machine learning analysis

“…The ML models considered have been previously evaluated and mathematically described, in conjunction with well log/lithofacies prediction datasets with a range of dependent variables. Examples of the published studies relevant to lithofacies predictions from well-log-derived input variables for each model evaluated include: ADA (Farzi and Bolandi, 2016;Wrona et al, 2018); DT and KNN (Wrona et al, 2018;Merembayev et al, 2021); MLP (Mandal and Rezzaee, 2019;Hossein et al, 2020); RF (Hossein et al, 2020;Merembayev et al, 2021); SVR (Mandal and Rezzaee, 2019;Hossein et al, 2020); and, XGB (Bestagini et al, 2017;Shashank and Mahapatra, 2018). The mentioned studies apply and compare the performance of several ML models.…”