摘 要:针对目前服装面料需求数据周期非确定性导致预测精度差的问题,提出一种基于 Prophet 算法的服装面料需求预测模型。通过分析服装面料需求历史数据的时间序列特征构建 Prophet 模型,使用 M 公司面料数据集设计 Prophet 与 LSTM 的对比实验,并采用 RMES 以及 MAE 作为评价指标。实验结果表明:相比于 LSTM,Prophet 模型具有较高的预测精度且有效提升了服装面料需求时间序列预测的准确性。
关键词:Prophet 模型;时间序列预测;LSTM;服装面料需求预测
DOI:10.19850/j.cnki.2096-4706.2021.20.024
基金项目:湖南省研究生科研创新项目(QL20210249)
中图分类号:TP18 文献标识码:A 文章编号:2096-4706(2021)20-0095-04
Research on Clothing Fabric Demand Forecast Based on Prophet Algorithm
LI Tingli, LI Changyun, WANG Songye
(Hunan Key Laboratory of Intelligent Information Perception and Processing Technology, Hunan University of Technology, Zhuzhou
412007, China)
Abstract: Aiming at the problem of poor prediction accuracy caused by the uncertainty of clothing fabric demand data cycle, a clothing fabric demand prediction model based on Prophet algorithm is proposed. By analyzing the time series characteristics of clothing fabric demand historical data, the Prophet model is constructed, the comparative experiment between Prophet and LSTM is designed by using the fabric data set of M company, and RMES and MAE are used as evaluation indexes. The experimental results show that compared with LSTM, Prophet model has higher prediction accuracy and effectively improves the accuracy of clothing fabric demand time series prediction.
Keywords: Prophet model; time series prediction; LSTM; clothing fabric demand forecast
参考文献:
[1] 江婷 .D 公司基于需求预测的运动服装面料库存优化研究 [D]. 上海:东华大学,2018.
[2] 董亚,杜景林,胡玉杰 . 基于 K-means 的模糊马尔科夫模型的降水预测 [J]. 现代电子技术,2021,44(19):85-89.
[3] GILBERT K. An ARIMA Supply Chain Model [J/OL].Management Science,2005,51(2):[2021-08-26].https://doi. org/10.1287/mnsc.1040.0308.
[4] 缪辉,唐晨添,罗露璐 . 基于 ARIMA 模型的新能源汽车销量预测 [J]. 企业科技与发展,2020(10):97-98.
[5] 常炳国,臧虹颖,廖春雷,等 . 基于选择性集成 ARMA 组合模型的零售业销量预测 [J]. 计算机测量与控制,2018,26(5):132-135.
[6] TAYLOR S J,LETHAM B. Forecasting at scale [J/OL].PeerJ Preprints,2017. [2021-08-26].https://doi.org/10.7287/peerj.preprints.3190v2.
[7] 苏新 . 基于 Prophet-LSTM 组合模型的服装销售预测研究 [D]. 镇江:江苏科技大学,2020.
作者简介:李亭立(1997—),女,汉族,湖南岳阳人,硕士在读,研究方向:工业大数据;李长云(1971—),男,汉族,湖南耒阳人,二级教授,博士,研究方向:软件理论、物联网工程、人工智能;王松烨(1996 -),男,汉族,河北石家庄人,硕士在读,研究方向:边缘计算。