A Novel Method for Predicting Sweet Spot Intervals of Shale Oil Based on CatBoost and Arithmetic Optimization Algorithm

The evaluation of sweet spot intervals (SSI) is a crucial aspect in the process of unconventional oil and gas exploration and development. However, traditional stacking methods struggle to quantitatively and reliably predict SSI due to the complexity of shale oil enrichment mechanisms. Therefore, this work proposed a novel method for predicting SSI based on categorical boosting (CatBoost) and arithmetic optimization algorithm (AOA). The continental shales of the Lower Permian Fengcheng Formation in the Mahu Sag, Junggar Basin were taken as the research object. A comprehensive and quantitative characterization was performed on the oil-bearing properties, reservoir properties, and compressibility. The optimal features for SSI classification were selected based on the geological significance of indicators and the Spearman correlation analysis. The CatBoost-AOA model was constructed to predict SSI by combining the dataset with petroleum geological theory. Moreover, the CatBoost algorithm was employed to address the missing data issues, and the 5-fold cross-validation was applied to ensure the objectivity and consistency of the model. The results indicated that the optimal features for classifying SSI were free hydrocarbon content, effective porosity, and brittleness index (BI). The model achieves precise SSI predictions with an accuracy of 0.9722, a precision of 0.9861, a recall of 0.9722, and an F1 score of 0.9756. It exhibits excellent performance in comparison with seven competing models on all performance evaluation metrics. By addressing the limitations of traditional petroleum geological interpretation techniques, this model enhances the efficiency and success rate of oil and gas exploration.