基于深度学习的快时尚服装产品销售预测模型构建
Construction of fast fashion clothing sales prediction model based on deep learning
李鑫 1胡永仕 1邵博 2苏晓丽3
作者信息
- 1. 福建理工大学 交通运输学院,福建 福州 350118
- 2. 美国威斯康星大学麦迪逊分校 工程学院,美国 麦迪逊 53706
- 3. 福州大学 经济与管理学院,福建 福州 350108
- 折叠
摘要
为了准确预测快时尚服装产品销售量,捕捉在间歇性或异常峰值销量中的时间信息,基于深度自回归模型,引入时间注意力机制,改进其网络结构设计,构建全局时序模型对快时尚服装产品销售进行预测.研究发现:基于注意力机制的深度自回归模型,能够从所有销售数据中有效学习到服装产品销售正常值与间歇性或异常峰值的时间关联关系,能够识别复杂模式下产品销售量的短期波动与长期趋势,且性能优于其他经典模型,验证了基于深度学习构建快时尚服装产品销售预测模型的可行性.
Abstract
In order to accurately predict the sales volume of fast fashion clothing products and capture the time information in the intermittent or abnormal peak sales volume,based on the deep autoregressive model,the time attention mechanism was introduced,the network structure design was improved,and the global timing model was constructed to predict the sales volume of fast fashion clothing products.The study found that the deep autoregressive model based on the attention mechanism can effectively learn the time correlation between the normal sales value of clothing products and the intermittent or abnormal peak value from all sales data,and can identify the long-term trend and short-term fluctuations of product sales under complex patterns,and its performance is better than other classical models.The feasibility of constructing fast fashion clothing sales forecasting model based on deep learning was verified.
关键词
深度学习/销售预测/数据驱动/快时尚/AT-DeepAR模型Key words
deep learning/sales prediction/data driven/fast fashion/AT-DeepAR model引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据