Application of LightGBM and LSTM combined model in vegetable sales forecast. (December 2020)

• Record Type: Journal Article
• Authors: He, Zhang; Yu, Sun
• Is Part Of: Journal of physics. Volume 1693(2020)
• Publication Date: 2020-12
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1693/1/012110
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: In view of the nonlinear and linear influence of vegetables in sales forecasting, the previous single model could not fully explore the variation law of vegetables in sales, a combined model based on LightGBM and LSTM is proposed, and the significant and abstract characteristics affecting sales forecasting are explored respectively by combining the advantages of the two models. First, the LightGBM model and the LSTM model based on intensive learning are modeled and analyzed, and then the two models are weighted array by the error reciprocal method for sales forecasting. The experimental results show that the proposed combination model based on LightGBM and LSTM is more accurate than the single model, and the prediction results of the model for vegetable short-term sales have provided important reference value for the strategic marketing of enterprises.