Manufacturing Big Data Modeling Algorithm Based on GM (1,1) - LSTM and Its Application in Sales Forecasting

• Published in: Data Driven Control and Learning Systems Conference (Online) pp. 1171 - 1175
• Main Authors: Long, Yinren, Xiao, Yi, Ren, Hongru, Lu, Renquan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 12.05.2023
• Subjects: Big Data; Big Data Modeling; Data models; Industries; Manufacturing Enterprise; Neural networks; Prediction algorithms; Predictive models; Resistance; Sales Forecasting
• ISSN: 2767-9861
• DOI: 10.1109/DDCLS58216.2023.10166094

It is a new period for the development of automobile industry, the economic situation is complex and changing, and the policies of automobile industry are frequently issued, so accurate prediction of automobile sales is extremely important and practical for both government and enterprises. In this paper, the GM(1,1) model and the long short-term memory (LSTM) neural network model are combined and optimized, and the sales of a brand of cars from January 2019 to September 2022 are used as sample data, and the car sales in the next three months are predicted by two single models and linear combination forecasting models, respectively. The experimental results show that the linear combined forecasting model outperforms the other two single models in terms of forecasting results and has better resistance to the interference of external factors.