基于LightGBM优化组合模型的销售预测
首发时间:2020-02-04
摘要:针对超市商品销量的预测问题,本文在研究大量文献的基础上,提出了一种基于LightGBM及XGBoost组合的预测模型。该模型不仅对商品的基本特征进行提取,同时结合了时间滑动窗口在数据特征处理上的优势,对商品销售数据进行动态特征提取,最后通过组合模型进行预测。结果显示,经过滑动窗口法进行数据特征提取后,最终组合模型的预测精度明显优于单模型的预测精度,实验表明,该模型对于超市商品的销量预测精度有明显的提升作用。
For information in English, please click here
Sales Forecast Based on LightGBM Optimized Combination Model
Abstract:Aiming at the prediction of supermarket sales volume, after studying various documents, a prediction model based on the combination of LightGBM and XGBoost is proposed in this paper according to the characteristics of various models. This model not only extracts the basic characteristics of commodities, but also combines the advantages of time sliding window in data feature processing to extract the dynamic characteristics of commodity sales data. Finally, make a prediction through a combination model. The results show that the prediction accuracy of the combined model is significantly better than that of the single model after the data feature extraction by the sliding window method. The experiment shows that the model can significantly improve the accuracy of the sales forecast of supermarket goods.
Keywords: Sales Prediction Feature engineering Combination Model LightGBM
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
基于LightGBM优化组合模型的销售预测
评论
全部评论0/1000