Abstract: Ventilated cavitating flow around an axisymmetric body involves coupled multi-parameter interactions with high experimental costs. A Gaussian process-based sequential sampling method was established to optimize experimental design and ensure the quality of the dataset while reducing the sample size. The effects of Froude number (Fr) and ventilation coefficient (C_Q) on cavity characteristics were investigated. In circulating water tunnel experiments, high-speed photography was used to capture cavity morphology under ventilation rates (5～85 L/min) and velocities (4～10 m/s). The results indicate that both increased C_Q and decreased Fr elongate cavities and modify their shapes, causing transitions from foam-like to re-entrant jet patterns with nonlinear coupling. The sequential sampling method dynamically adjusted the sample distribution, achieving <5% prediction error with Gaussian process modeling, and outperformed static sampling in data efficiency. Additionally, a cavity length distribution model was developed to classify characteristic variations across length types. The findings offer a practical approach for optimizing ventilated cavitation in engineering applications and serve as a reference for designing simulation experiments for nonlinear and complex flow in future research.