[1]卢晨辉,冯硕,易爱华,等.基于深度学习的加油站销量预测与营销策略应用研究[J].郑州大学学报(工学版),2022,43(01):1-6.[doi:10.13705/j.issn.1671-6833.2022.01.014]
 Lu Chenhui,Feng Shuo,Yi Aihua,et al.A Deep Learning ba<x>sed Gasoline Station Sales Prediction and its Application on Promotion Strategy[J].Journal of Zhengzhou University (Engineering Science),2022,43(01):1-6.[doi:10.13705/j.issn.1671-6833.2022.01.014]
点击复制

基于深度学习的加油站销量预测与营销策略应用研究()
分享到:

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
43卷
期数:
2022年01期
页码:
1-6
栏目:
出版日期:
2022-01-09

文章信息/Info

Title:
A Deep Learning ba<x>sed Gasoline Station Sales Prediction and its Application on Promotion Strategy
作者:
卢晨辉冯硕易爱华叶晓俊
清华大学软件学院;中石化销售股份有限公司广东石油分公司;

Author(s):
Lu Chenhui; Feng Shuo; Yi Aihua; Ye Xiaojun;
College of Software of Tsinghua University; Guangdong Petroleum Branch of Sinopec Sales Co., Ltd.;

关键词:
Keywords:
DOI:
10.13705/j.issn.1671-6833.2022.01.014
文献标志码:
A
摘要:
营销策略的制定是加油站业务的重要部分,而数据驱动的营销策略制定已成为加油站实现精准营销的迫切需求。本文提出了一种基于加油站历史数据、营销策略和关键特征的油品销量预测的深度学习模型和基于销量预测模型的营销策略制定方法。根据加油站历史数据特征,我们设计了一个多层次的网络结构处理不同类别特征的数据,并结合营销策略信息以执行油品的销量预测。此外,通过引入关键特征,我们提升了销量预测模型的准确度;通过输入营销策略信息的变更,我们实现了加油站营销策略的自动选择。在真实加油站数据构建的数据集上进行的实验的结果显示,我们提出的销量预测模型相比其他主流方法具有更低的预测误差。
Abstract:
Promotion strategy formulation is an important part of gas station business, and data-driven promotion strategy formulation has become an urgent demand for gas stations to achieve precise marketing. This paper proposes a deep learning model for forecasting gasoline sales ba<x>sed on historical gas station data, promotion strategies and key features, and a promotion strategy formulation method ba<x>sed on sales forecasting models. Due to the historical data characteristics of gas stations, we design a multi-level network structure to process data of different types, and combine promotion strategy information to perform oil sales forecasts. In addition, by introducing key features, we have improved the accuracy of the sales forecast model by inputting different promotion strategies, we have realized the automatic selection of gas station marketing strategies. The results of experiments conducted on a data set constructed from real gas station data show that the sales forecast model we proposed has lower forecast errors than other mainstream methods
更新日期/Last Update: 2022-01-09