<![CDATA[Support Vector Machine (SVM) algorithm is a machine learning method renowned for its high accuracy and computational efficiency in prediction and classification tasks. In this study, SVM was applied to predict effective porosity from well log data. The prediction model was optimized using GridsearhCV (GS CV) module and tested on seven wells from the 'Mentari' field, Indonesia. Six variations of training-testing configuration were evaluated to assess the prediction performance. The best ere acjieved using four training wells and three testing wells, yielding an accuracy of 71% with training time of 1.98 seconds. The analysis revealed that increasing the volume of training data improves accuracy, albeit with longer computational time. This study confirms that SVM demonstrates strong predictive capability for effective porosity and has the potential to serve as a supporting tool in simplifying geological interpretation, particularly during the initial analysis stage when sufficient training data is available.]]>
Copyrights © 2025
