One of the important reservoir issues is the estimation of its petrophysical parameters. The volume of shale in the formation affects the petrophysical parameters of the reservoir such as porosity, permeability, and saturation. Since these parameters indicate the amount of hydrocarbon in the reservoir and the ability to produce them, shale volume determination is an essential challenge in the oil industry. This study compares the conventional methods and Machine Learning (ML) algorithms to estimate the shale volume in the Kazdumi Formation in one of the oil fields in southwest Iran. This formation has shaly-sand lithology and due to the complex conditions in the sedimentation environment, the distribution of its shale horizons is completely heterogeneous. In this regard, the shale volume of the formation was calculated using the Gamma-Ray (GR) method, and the results were validated with core data. After that, conventional methods including Gamma-Ray, SP, Neutron, Density, Sonic and Resistivity and also Density-Neutron, and Density-Sonic as well as ML methods such as Artificial Neural Networks (ANN), Support Vector Machine (SVM), and Random Forest (RF), Least Square Boost(LSBoost), Bayesian and Linear Regression methods have been used to calculate shale volume of Fahliyan Formation. RXOZ, RXO, RT, DT, NPHI, RHOZ, SP, HCAL and PEFZ logs are used as input data in these methods. The accuracy of each method was measured using statistical parameters including correlation coefficient (R2) and Mean Square Error (MSE). Results show that the conventional methods have low accuracy and not an acceptable performance while machine learning algorithms have more reliable results and are technically optimal. Among the studied methods, RF with 98% accuracy performed better than other algorithms in estimating reservoir shale volume.