Abstarct: The volume of shale is a pivotal parameter in petrophysical analysis that allows the precise estimation of other petrophysical parameters such as effective porosity, and water saturation. This is a crucial step in the characterization of reservoirs as well as evaluation of hydrocarbon potentials. This study compares classical methods with machine learning algorithms to estimate the shale volume in one of iranian oil fields in the north of the Persian Gulf. The data used in this study were collected baxsed on real data points acquired from two wells. These data include core and well log data, such as gamma-ray, sonic, density, neutron, and resistivity logs. The dataset from well 1 with 2395 data points, was utilized for training and testing the models. These data were split into 80% for training and 20% for testing. Additionally, the dataset from well 4 with 1738 data points, was utilized to evaluate the models. After data preprocessing, four features including sonic, density, neutron, and laterolog deep (resistivity) were used as input to estimate shale volume. The sensitivity analysis revealed that an artificial neural network with 1 hidden laxyer, 24 neurons, ReLU as the activation function, and Adam as the training function, and random forest with 300 trees, maximum tree depth of 10, minimum samples required to split a node set to 5, and minimum samples required in a leaf node set to 4, are the best models for estimating shale volume. The shale volume in the validation well data (4) was estimated using classical methods such as neutron-density, sonic-density, and neutron with coefficients of determination (R^2) of 0.85, 0.84, and 0.82, and the root mean square error (RMSE) of 0.119, 0.124, and 0.127, respectively. The results showed, the neutron-density method demonstrated better performance compared to other methods. Furthermore, the shale volume in the validation well data (4) was estimated using an artificial neural network, and random forest methods, with coefficients of determination of 0.92 and 0.89, and the root mean square error of 0.088, and 0.105 respectively. The results showed, the artificial neural network method outperforms the random forest method. Therefore, in this study, the artificial neural network model demonstrated better performance in estimating the volume of shale, with a higher coefficient of determination and lower root mean square error compared to all evaluated methods.