基于多源数据的品牌汽车需求预测研究

发布时间：2018-03-04 08:16

本文选题：多源数据　切入点：供应链管理　出处：《合肥工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：经济全球化和改革开放推动了我国汽车工业的发展,促使新经济时代来临。在新经济时代,居民的消费由“住房消费”转向“汽车消费”,市场竞争也演化成供应链之间的竞争。在以需求驱动为主的供应链管理模式下,需求成为供应链的起点和动力源泉。实时、精准的需求预测不仅有利于消费者合理选择购买时机,也成为供应链管理模式下的战略性问题。传统模式下,消费者在进行购买决策调研过程中主要通过自身经验、公开信息、逛商场等方式进行,供应链的需求信息在传递过程中具有一定的时延和扭曲。这种模式下的需求预测方法由于数据信息匮乏、实时性较差,因此其预测准确率也较低,牛鞭效应的现象在供应链中也经常存在,影响供应链整体的效率。互联网的深入发展使得消费者可以很方便地在网上进行购前调研,网上行为成为消费者购买决策过程中的重要环节。这使得捕获多源的大数据来进行预测成为可能,同时也带来了一些数据选择和融合方面的挑战。基于当前背景,为应对数据选择和融合方面的挑战,本文以多源数据为基础,对供应链管理模式下的品牌汽车需求预测进行研究。首先,提出一种基于多源数据信息共享供应链协同预测框架,对其运作流程和层次划分进行论证分析。然后,进一步提出基于多源数据的预测模型方法。将获取的多源数据按照汽车品牌进行分类、标注、关联和统计,提取出数据指标。通过因素选择模型来分析、筛选关联度较大的数据指标作为模型的输入,结合选择后的多源数据指标,综合考虑时间序列和多种影响因素来进行品牌汽车需求预测研究。本文的研究成果,针对新背景下的数据选择和融合问题,考虑到理论和实践意义,提出了解决方案,为供应链管理模式下的汽车需求预测提供了新的思路,对需求预测的研究具有一定积极意义。
[Abstract]:Economic globalization and reform and opening to the outside world have promoted the development of China's automobile industry and promoted the advent of a new economic era. Residents' consumption has changed from "housing consumption" to "automobile consumption", and market competition has evolved into competition between supply chains. In the demand-driven supply chain management mode, demand becomes the starting point and power source of supply chain. Accurate demand forecasting is not only helpful for consumers to choose the right time to purchase, but also becomes a strategic problem in the mode of supply chain management. In the traditional mode, consumers mainly disclose information through their own experience in the process of purchasing decision research. The demand information in the supply chain has some delay and distortion in the transmission process. Because of the lack of data information and the poor real-time performance, the forecasting accuracy of the demand prediction method in this mode is also low. Bullwhip effect often exists in the supply chain, which affects the overall efficiency of the supply chain. The further development of the Internet makes it convenient for consumers to conduct pre-purchase research on the Internet. Online behavior has become an important part of consumer buying decisions. This has made it possible to capture multiple sources of big data to make predictions, and it has also brought some data selection and fusion challenges. In order to meet the challenges of data selection and integration, this paper studies the demand forecasting of brand cars based on multi-source data. Firstly, a collaborative forecasting framework for supply chain based on multi-source data sharing is proposed. Then, a prediction model method based on multi-source data is put forward. The obtained multi-source data is classified, labeled, correlated and counted according to the automobile brand. The data index is extracted and analyzed by the factor selection model, and the data index with high correlation degree is selected as the input of the model, and the selected multi-source data index is combined with the selected multi-source data index. The research results of this paper, aiming at the problem of data selection and fusion under the new background, considering the theoretical and practical significance, put forward a solution. It provides a new idea for automobile demand forecasting under the supply chain management mode, and has certain positive significance to the research of demand forecasting.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F426.471

【参考文献】