An Empirical Analysis of Anhui Province's Quarterly GDP based on the LSTM and SARIMA Combined Model

Kai Chen

doi:10.62051/ijcsit.v3n3.47

Authors

Kai Chen

DOI:

https://doi.org/10.62051/ijcsit.v3n3.47

Keywords:

LSTM neural network, SARIMA model, Combination prediction, Natural selection particle swarm algorithm

Abstract

This paper uses the data of Anhui Province's quarterly GDP from 2005 to 2022 to build different models for prediction. Two single models, LSTM neural network model and SARIMA model, are established for prediction respectively. On this basis, the combination weights are determined by particle swarm optimization algorithm. It is found that the mean square error (MSE) of the combination using PSO particle swarm optimization algorithm is 0.2236 and 7.0159 lower than that of the two single prediction models, respectively. That is, the proposed combination model based on natural selection particle swarm algorithm to determine the weights of LSTM neural network model and SARIMA model has better prediction effect and higher prediction accuracy. This combination prediction model can be considered to predict the future GDP data of Anhui Province.

Downloads

Download data is not yet available.

References

[1] Zhang Shuhong, Yang Wancai, Wu Xingan. Forecast of per capita GDP in Henan Province during the "Twelfth Five-Year Plan" period [J]. Mathematical Statistics and Management, 2014, 33(03): 394-399. DOI: 10.13860/j.cnki.sltj. 2014.03.003.

[2] Hong Yan, Liu Bingxiang, Wan Xiang. A GDP forecasting method based on interpolation fitting [J]. China Management Informatization, 2016, 19(16): 123-124.

[3] Zhang Jing. Forecast of per capita GDP in Gansu Province based on Bayesian time series model [J]. Mathematical Statistics and Management, 2018, 37(02):205-210. DOI:10.13860/j.cnki.sltj.20170721- 001.

[4] Tian Zichen, Liu Miao. Empirical GDP forecasting based on the improved gray GM(1,1) model[J]. Statistics and Decision Making, 2018, 34(11):83-85. DOI:10.13546/j.cnki.tjyjc .2018.11.019.

[5] Wang Shasha, Chen An, Su Jing, et al. Application of combination forecast model in China's GDP forecast [J]. Journal of Shandong University (Science Edition), 2009, 44(02): 56-59.

[6] Zhang Zhichao. Application of combination forecast model in per capita GDP forecast of Anhui Province [J]. Journal of Guangxi University of Finance and Economics, 2009, 22(05): 32-35.

[7] Xiong Zhibin. Research on GDP time series forecasting based on the integration of ARIMA and neural network[J]. Mathematical Statistics and Management, 2011, 30(02):306-314. DOI:10.13860/j.cnki.sltj.2011.02.012 .

[8] Long Huidian, Yan Guangle. Research on GDP time series forecasting based on SARIMA, GM(1,1) and BP neural network integrated model [J]. Mathematical Statistics and Management, 2013, 32(05): 814-822. DOI:10.13860/j.cnki.sltj.2013.05.004.

[9] Mo Songjuan, Yang Guiyuan, Luo Yang. Combination forecast of China's total GDP based on IOWGA operator [J]. Statistics and Decision Making, 2015, No. 436(16): 4-7. DOI: 10.13546/j. cnki.tjyjc.2015.16.001.

[10] Li Yanfei, Wu Tao, Guo Haiyan. Anhui Province per capita GDP combination forecast model based on proximity [J]. Journal of Chongqing Technology and Business University (Natural Science Edition), 2019, 36(04):95-100. DOI:10.16055/j .issn.1672-058X.2019.0004.016.

[11] Li Jia, Huang Zhihao, Wang Jiahui. GDP forecast based on independent recurrent neural network method [J]. Statistics and Decision Making, 2020, 36(14): 24-28. DOI: 10.13546/j.cnki.tjyjc.2020.14. 005.

[12] Liu Sifeng. The emergence and development of gray system theory [J]. Journal of Nanjing University of Aeronautics and Astronautics, 2004(02):267-272. DOI:10.16356/j.1005-2615.2004.02.027.

[13] Li Wan, Feng Fenling, Jiang Qiwei. Railway passenger volume prediction using improved particle swarm algorithm to optimize LSTM neural network [J]. Transactions of Railway Science and Engineering, 2018, 15(12): 3274-3280. DOI: 10.19713/j. cnki.43-1423/u.2018.12.033.

[14] Hu Jianbo, Luo Zhipeng, Li Feng. Forecasting China’s carbon emission intensity under the “carbon peak” goal—analysis based on LSTM and ARIMA-BP models [J]. Financial Science, 2022, No. 407(02):89- 101.

[15] Gu Xin. China's economic development forecast based on ARIMA-LSTM-XGBoost combination model [D]. Qingdao University, 2021. DOI:10.27262/d.cnki.gqdau.2021.001857.

[16] Ma Jingwen, Li Shuqing, Xia Mengyao. Review of research on the application of machine learning in GDP forecasting [J]. Science and Technology Information Research, 2022, 4(03):73-94. DOI:10.19809/j.cnki.kjqbyj.2022.03. 008.

[17] Wang Jiawei, Liu Miao, Tian Zichen. Empirical analysis and prediction of Xinjiang’s GDP based on improved gray model [J]. Practice and Understanding of Mathematics, 2019, 49(17): 306-312.

[18] Chen Ting. Comparative study on the application of ARIMA model and BP neural network model in AIDS incidence prediction [D]. Guangxi Medical University, 2015.

[19] Li Yongdi. Forecast of total GDP and growth rate based on ARIMA, deflator method and principal component regression model - taking Henan Province as an example [J]. Henan Science and Technology, 2021, 40(13): 149-155.