A method of forecasting trade export volume based on back-propagation neural network

Area of Science:

• Economics
• Data Science
• Financial Modeling

Background:

• Financial forecasting accuracy decreases at long horizons.
• Foreign trade is crucial for economic growth, technology transfer, and employment.
• Traditional forecasting methods struggle with the nonlinear dynamics of foreign trade.

Purpose of the Study:

• To predict foreign trade export volume using a back-propagation neural network (BPNN).
• To evaluate the effectiveness of BPNN for nonlinear financial forecasting.
• To compare BPNN model accuracy against traditional forecasting techniques.

Main Methods:

• Utilized a back-propagation neural network (BPNN) for export volume prediction.
• Developed both multifactor and single-factor forecasting models.
• Compared forecasted export volumes with actual data for a Chinese city.

Main Results:

• The BPNN model demonstrated over 30% higher forecasting accuracy compared to traditional methods.
• The application of the BPNN model showed approximately 15% greater accuracy.
• The model's predictions closely aligned with the growth trends of actual export data.

Conclusions:

• Back-propagation neural networks are highly effective for nonlinear financial forecasting.
• BPNN offers a superior approach to predicting foreign trade export volumes.
• Developing foreign trade is a vital strategy for driving economic growth.