Adaptive Machine Learning Methods in Oil Well Research

This article discusses the oil well exploration model using neural networks and prediction in Python. The problem of identifying significant features is considered. Decision tree algorithms, Sequential Forward Selection (SFS), Sequential Backward Selection (SBS) and Exhaustive Feature Selection (EFS) are considered as selection methods. Graphs are presented illustrating comparisons of methods and results of algorithms, and an assessment of the quality of methods is also considered. The paper considers several types of neural networks for building an oil well survey model based on learning methods: the Levenberg-Marquardt method, the Bayesian regularization method, the error backpropagation method. The results of evaluating effective performance indicators from the above teaching methods and methods for highlighting significant features are shown.