This article discusses the process of predicting failures of deep-pumping equipment in the oil industry. A procedure for predicting failures of oil well equipment is proposed, which is based on the application of a gradient boosting algorithm and includes three stages. At the first stage, a set of features necessary for the analysis is formed, as well as a sample of data. At the second stage of data preprocessing, the following are performed: processing of missing values; coding of categorical features; getting rid of outliers by the interquartile range method. At the third stage, the gradient boosting algorithm is applied to predict failures of deeppumping equipment. The peculiarity of the algorithm implementation consists in the proposed scheme of actions, which means sequential fixing first the number of decision trees, and then the learning rate, involving the analysis of the dependencies of the root-mean-square error on both the learning rate and the number of trees. The result of predicting failures of deep-pumping equipment of an oil well is presented and the results of experimental studies using the developed models on real data are described. The predicted value of the number of failures of the deep-pumping equipment of the oil well for the subsidiary of the company Rosneft Nyaganneftegaz was calculated. The proposed procedure for solving the problem of predicting equipment failures based on the application of the gradient boosting algorithm is advisable to use in the process of obtaining an engineering education both when studying special disciplines and when performing final qualifying works.