Under the background of complex and changeable economic environment and massive data growth, this article aims to build a more accurate and effective econometric forecasting model. Firstly, this article expounds relevant theories such as econometrics, deep learning (DL) and big data technology, and analyzes the limitations of traditional econometric forecasting methods. Then, based on the integration of DL and big data, through data collection, cleaning, feature engineering and other big data processing processes, the back propagation neural network (BPNN) algorithm is selected and improved to build the model architecture. Through experiments, based on the data of GDP growth rate, inflation rate and enterprise revenue, the linear regression model (LR), autoregressive integrated moving average model (ARIMA) and simple neural network (SNN) are compared. The results show that the new model performs well in the mean square error (MSE), mean absolute error (MAE) and mean absolute percentage error (MAPE). In the prediction of GDP growth rate, the MSE of the new model is stable at around 0.015, which is much lower than other comparative models. This shows that DL and big data-driven econometric forecasting model have obvious advantages in accuracy, stability and robustness, and can provide strong support for economic decision-making.
Haochen Guo (Sun,) studied this question.