Regional economic forecast using Elman neural networks with wavelet function.

Recently, the economy in Guangdong province has ranked first in the country, maintaining a good growth momentum. The prediction of Gross Domestic Product (GDP) for Guangdong province is an important issue. Through predicting the GDP, it is possible to analyze whether the economy in Guangdong province can maintain high-quality growth. Hence, to accurately forecast the economy in Guangdong, this paper proposed an Elman neural network combining with wavelet function. The wavelet function not only stimulates the forecast ability of Elman neural network, but also improves the convergence speed of Elman neural network. Experimental results indicate that our model has good forecast ability of regional economy, and the forecast accuracy reach 0.971. In terms of forecast precision and errors, our model defeats the competitors. Moreover, our model gains advanced forecast results to both individual economic indicator and multiple economic indicators. This means that our model is independently of specific scenarios in regional economic forecast. We also find that the investment in education has a major positive impact on regional economic development in Guangdong province, and the both surges positive correlation. Experimental results also show that our model does not exhibit exponential training time with the augmenting of data volume. Consequently, we propose that our model is suitable for the prediction of large-scale datasets. Additionally, we demonstrate that using wavelet function gains more profits than using complex network architectures in forecast accuracy and training cost. Moreover, using wavelet function can simplify the designs of complexity network architectures, reducing the training parameter of neural networks.
Competing Interests: The authors have declared that no competing interests exist.
(Copyright: © 2024 Liang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.)