Prediction of the Effective Acceleration-to-Velocity (EPA/EPV) Ratio Using Hyperparameter-Optimized and Physically Interpretable Machine Learning Models Based on Regional Seismic Data