网络搜索指数与汽车销量关系研究.doc

网络搜索指数与汽车销量关系研究 　　〔摘要〕网络搜索数据是消费者在信息搜集和购买决策过程中真实足迹的反映，对了解消费者购买需求具有重要价值。本文运用与现有研究不同的关键词获取方法，以我国汽车市场为背景，研究网络搜索数据与销量之间的关系。首先，确定网络搜索数据的关键词，主要运用了文本挖掘技术，具体而言：①对抓取的汽车论坛文本进行Jieba分词；②利用Word2vec模型把分词结果转化为向量空间模型形式；③结合TF-IDF算法和余弦相似度算法确定关键词。然后，基于108个月的长面板数据，建立网络搜索与汽车销量的固定效应模型。最后，采取滚动窗口的方式预测最近12个月的汽车销量。实证结果显示：网络搜索与汽车销量之间存在长期均衡关系，回归模型可以解释76%的方差；网络搜索数据有助于预测我国汽车销量 〔关键词〕网络搜索数据；消费者；购买需求；汽车销量；文本挖掘；关键词获取；长面板数据；预测 DOI：10.3969/j.issn.1008-0821.2016.08.026 〔中图分类号〕F27213〔文献标识码〕A〔文章编号〕1008-0821（2016）08-0131-06 〔Abstract〕The online searching data reflects the real traces that consumers leave behind while gathering some information or making purchase decisions.The online searching data benefits us to know the information of what is the consumers need.In our paper，the selection of keywords differents from the existing researches，on the background of Chinas automobile market，the paper researched the relationship between online searching data and Chinese automobile sales.Firstly，identified the key words of online searching mainly according the text mining technology，to be specific：①Using Jieba segmenting the automobile BBS text captured to words；②Converting segmentary words to the Vector space model by Word2vec model；③Combining TF-IDF algorithm and cosine similarity algorithm to determinate the key words.Then，based on the 108 months long panel data，the paper established a fixed effect model between the online searching data and the automobile sales.Finally，the paper forecasted the automobile sales in nearly 12 months.The empirical result showed that：there is a Long-run equilibrium relationship between the online searching data and the automobile sales，and regression model can explain 76% of the variance.The holdout analysis suggested that online searching data can be of substantial use to forecast the Chinas automobile sales. 〔Key words〕online searching data；consumer；purchase decision；automobile sales；text mining；keywords obtaining；long panel data；forcast 搜索引擎已经成为消费者决策过程中信息搜索和评价过程