The formulation of oilfield producting engineering project designs is a complex decision-making process. Traditional empirical methods are no longer sufficient to meet scientific requirements. For predictable decision-making problems, such as the prediction of pumping unit system efficiency, big data analysis methods can be employed to assist in making designs more scientific. In this study, a precise and stable pumping unit system efficiency prediction model is developed using the random forest algorithm. By integrating a large amount of pumping unit operational data, including pump parameters, downhole conditions, surface equipment status, and operating parameters, among other diverse influencing factors, thorough data preprocessing, feature engineering, and model training are conducted. Through methods such as grid search, random search, and Bayesian optimization, the parameters of the random forest model are finely tuned to address the common issue of overfitting associated with applying the random forest algorithm. Additionally, by selecting an appropriate number of decision trees and features to improve the model’s predictive performance, a random forest model with strong generalization ability and effective data pattern capture is constructed. The model, rigorously validated, demonstrates superior predictive performance compared to traditional methods, enriching the decision-making methods for oilfield producting engineering project designs and providing strong support for optimizing operational parameters, reducing energy consumption, and mitigating carbon emissions.