Predicting the spatial distribution of hydrocarbon resources is a critical task in oilfield exploration and development. To enhance the efficiency and accuracy of oil and gas distribution prediction, a novel Bayesian network classifier (BNC) is introduced to estimate the spatial distribution of hydrocarbon resources. First, a k-dependent Bayesian classifier based on the mutual information contribution rate (MSKDB) is proposed to address the limitations of current popular methods. Subsequently, considering the reservoir of the first member of the Shahejie Formation ( \(Es_1\) ) in southeastern Jizhong Depression in the Bohai Bay Basin as a case study, and based on the systematic analysis and extraction of key controlling factors, the MSKDB model is constructed and compared with other methods including the tree-augmented naïve Bayes (TAN), logistic regression (LR), and averaged one-dependence estimator method (AODE). The effectiveness and superiority of the new method are qualitatively and quantitatively analysed. Finally, according to the prediction results of the MSKDB method, the distribution range of the remaining hydrocarbon resources is intuitively displayed, and two categories of favourable zones for future exploration are preferentially selected, namely, types A (outstepping exploration region) and B (extension region). The results of the case study demonstrate that the proposed approach can effectively integrate the key controlling factors of the spatial distribution of hydrocarbons, accurately predict the distribution pattern of the remaining hydrocarbon resources in the Shahejie Formation, and provide a qualitative and quantitative basis for the next phase of exploration and deployment.
