Abstract: FA-RBF neural network algorithm is proposed by combining factor analysis algorithm, which has advantage of reducing the data's dimensions, with RBF neural network.Using the new algorithm, the original data can first reduce its dimension, then, the processed data can be the input of neural network to network training and simulation.FA-RBF algorithm can effectively simplify the network structure, improve the convergence speed and save training time.Apply the algorithm for the prediction of private car ownership, the prediction results show that compared to PCA-RBF and RBF neural network algorithm FA-RBF algorithm significantly improved prediction accuracy, reduced training time and the sum of squared errors.