Abstract: The dynamic response of the seat-occupant system to vibration exhibits complex dependencies on various factors, including anthropometric parameters, the seat structure, and riding environments, etc. Drawing from an experimental investigation into whole body vibration, this study aims to develop a predictive model for seat transmissibility using support vector machine regression. The recursive feature elimination method and principal component analysis are employed to reduce the dimensionality of anthropometric parameters. The resultant low-dimensional features are then integrated into the model to predict the seat transmissibility and cross-axis effect. The results show that compared with the traditional support vector machine regression model, the application of principal component analysis to reduce the correlation of anthropometric parameters can significantly reduce the prediction error of the model, and the fitting degree between the predicted value and the measured value can reach 92%. The accuracy of prediction can be further improved by removing the minor anthropometric parameters through the recursive feature elimination method, and the fitting degree between the predicted value and the measured value is 94%. The findings demonstrate that the support vector machine regression model, based on the feature dimensionality reduction optimization, can effectively sieves out the redundant information of input parameters in the model. This refinement enhances both the efficiency and accuracy of predicting the seat transmissibility.