This paper presents a study of the classification of igneous rocks in Eastern Oregon using a neural network. The main goal of the study was to create a model capable of classifying rocks based on their geochemical composition, such as the content of oxides SiO2, Al2O3, FeO, CaO and others. To complete the task, we used machine learning methods that include a neural network trained on data on basalts, andesites, and rhyolites. As a result, the classification accuracy was achieved at the level of 80%. Analysis of the importance of these features has shown that the content of calcium, magnesium, and silicon oxide is the most important factor for accurate identification of rocks. The results of the study demonstrate the effectiveness of using neural networks to analyze geochemical data and allow us to better understand the processes of igneous rock formation on a geological scale.