Research on Revenue Prediction of Automobile Manufacturing Enterprises

This paper establishes a quarterly revenue prediction model for automotive manufacturing enterprises. The whole model is divided into two parts. In the first part, a model based on Random Forest and SARIMA is established to predict the sales volume of the automobile industry. Results show that when SARIMA adopts static prediction, the RMSE and MAE of this method are 22.48% and 24% lower than those of the benchmark model that only relies on SARIMA to predict the sales volume. When SARIMA adopts the dynamic prediction method, the proposed method reduce RMSE and MAE by 17.36% and 19.11%, respectively, compared to the benchmark model. Then the predicted results are introduced into the second experiment, and a model for predicting enterprises quarterly revenue is constructed based on CNN, combining industry sales and financial features. The results show that the RMSE and MAE of quarterly revenue prediction method proposed reduce 63.32% and 0.07% respectively compared with the benchmark model based on the Random Forest algorithm.