Design/methodology/approach Seal test was carried out using a specialized seal test rig. Various data points were collected during this test, including leakage, friction torque, oil temperature, oil pressure and rotating speed. The collected data underwent noise separation and reconstruction using the complete ensemble empirical mode decomposition with adaptive noise method. Subsequently, a leakage prediction model is developed using the random forest regression with parameter optimization. A quantitative evaluation for influencing factors in leakage prediction process is investigated. Findings The results achieve a mean accuracy index exceeding 95%, demonstrating close alignment between predicted and actual leakage values. Feature contribution results highlight that the trends of the oil temperature, friction torque and oil pressure significantly affect the leakage prediction, with the oil temperature trend exerting the most substantial influence. Originality/value This work sheds light on the interplay between operating conditions and sealing performance degradation, offering valuable insights for understanding and addressing sealing issues effectively. Peer review The peer review history for this article is available at: https://publons.com/publon/10.1108/ILT-07-2024-0271/