Construction and Application of an MR-BP Neural Network Combined Forecasting Model Based on Adaptive Weighted Fusion

DOI: https://doi.org/10.62517/jbdc.202601223

Author(s)

Yao Wang, Zhixiang Yang*

Affiliation(s)

School of Mathematics and Data Science, Changji College, Changji, Xinjiang, China *Corresponding Author

Abstract

In view of the deficiency of traditional multiple regression model in dealing with nonlinear relations, and the poor interpretability and high training cost of BP neural network model, this paper proposes a combined forecasting model combining multiple regression and BP neural network, which takes into account the advantages of linear trend catching and nonlinear relationship mining. Based on the data of GDP per capita and related macroeconomic indicators of China from 2005 to 2023, this paper constructs a fusion framework to co-optimize the analytical ability of multiple regression and the fitting ability of neural network. The results show that the combination model proposed in this paper can fit the complex trend of GDP per capita more accurately, and is significantly better than other comparative models in prediction accuracy and stability, and effectively reduces the prediction error.

Keywords

Multiple Regression; BP Neural Network; Combined Model; GDP Forecasting

References

[1]Zhang Xiaoguo, Li Shuang, Li Yajie. Grey Relational Analysis of GDP Forecasting and Industrial Structure in Henan Province. China Market, 2025, (28): 12-16. [2]Liang Peiting, Wang Binhui, Wang Guochang, et al. Stock Price Forecasting Based on Automatic Identification of Linear or Nonlinear Structures. Journal of Systems Engineering, 2025, 40(04): 574-594. [3]Qiu Dong, Wang Zhongfei, Zhang Suixing, et al. User Electricity Demand Forecasting Method Based on BP Neural Network. Industrial Control Computer, 2025, 38(10): 111-112. [4]Wu Tingting. Prediction Model of Shoulder-Hand Syndrome in Elderly Stroke Patients Based on Lasso-Logistic Regression Analysis. Chinese Journal of Hospital Statistics, 2025, 32(04): 255-262. [5]Yao Ting, Li Hua, Wang Yujiang. Multiple Regression Model of Concrete Adiabatic Temperature Rise Based on Statistical Analysis. Concrete, 2025, (07): 51-54+67. [6]Xu Min, Lin Weixing, Shi Lei, et al. Prediction of Blasting Vibration in Mining Areas Based on I-GWO-BP Neural Network. Mining Research and Development, 2025, 45(10): 121-128. [7]Wang Wei, Cai Zhongyuan, Chen Jiongzhen. Compressor Performance Prediction Model Based on Multiple Linear Regression. Gas Turbine Technology, 2025, 38(03): 37-42. [8]Liu Lina, Song Qixiu, Zhang Lun, et al. Lasso-Logistic Regression Analysis of Direct Oral Anticoagulant-Related Issues and Construction of a Nomogram Prediction Model in Real-World Settings. Practical Pharmacy and Clinical Medicine, 2025, 28(06): 406-412. [9]Feng Jianwen, Wu Yuzhi. Construction of a Risk Prediction Model for Rheumatoid Arthritis-Associated Interstitial Lung Disease Based on Lasso-Logistic Regression Analysis. Journal of Clinical Pulmonary Medicine, 2024, 29(11): 1638-1643. [10]Zhang Yali, Chen Binha, Jin Ying. Construction of a Ulcer Recurrence Prediction Model for Patients with Lower Extremity Venous Ulcers Based on Logistic Regression Analysis. Chongqing Medicine, 2025, 54(07): 1576-1580. [11]Liu Xuexiang, Cui Yangyang, Zhu Hailu. Research on Short-Term Power Load Forecasting Based on an Improved RBF Neural Network. Electrical Technology and Economy, 2025, (10): 4-6. Qin Yanjiao. Comparative Study of House Price Prediction Models Based on Multiple Linear Regression and Random Forest Algorithm. Modern Information Technology, 2024, 8(22): 127-13