百度指数与旅游景区游客量的关系及预测研究——以北京故宫为例

网络搜索数据记录了用户的搜索关注与需求，为研究旅游经济行为提供了必要数据基础。文章基于百度指数，以北京故宫为例，利用计量经济学中的协整理论和格兰杰因果关系分析了百度关键词与北京故宫实际游客量间的关系，建立了没有百度关键词和加入百度关键词的两种预测模型并进行了预测精度比较。结果表明：故宫实际游客量与百度关键词存在长期均衡关系和格兰杰因果关系：加入百度关键词后的自回归分布滞后模型的样本期内的预测精度比没有百度关键词的ARMA模型提高了12．4％，样本期外的预测精度提高了14．5％。运用带有百度关键词的模型可以实现利用当天及滞后1～2天的百度指数数据预测故宫当天的游客量，不仅增强了预测的时效性，还可以更加及时、准确地为故宫景区管理部门提供决策的依据。

Study on the Predictive and Relationship between Tourist Attractions and the Baidu Index.A Case Study of the Forbidden City

Tourists overflowing during the ＂Golden Week＂ is not an uncommon situation in China today. Predicting tourist flows is significant for tourist attractions management and planning. Most existing methods rely on well- structured statistical data published by the government. This approach is limited in two aspects： 1 ） there may be significant delay in the predication, since governmentally published data are usually hysteretic; 2） the sample size can be small, leading to inaccurate prediction results. Recently, researchers in the economic and management domains have started to use internet search engines as data collecting tools for economic behavior monitoring and prediction. Internet search records can reflect concerns and interests of potential tourists, and provide a large volume of unstructured or semi-structured data for studying tourism economic behavior. This paper proposes a novel approach for predicting tourist flow based on the Baidu Index. Baidu is the global leading Chinese search engine. The Baidu Index provides search history containing different keywords on a daily basis dating back to 2006. In this paper, we conduct a case study using search data related to the Forbidden City from the Baidu Index and statistical data of tourist flows in the Forbidden City. The presented approach uses the econometric cointegration theory and Granger causality analysis to find relationships between the internet search data and the actual tourist flow. The paper compares analysis results obtained by two kinds of predictive models with or without considering Baidu Index. The study shows that there is a long-term equilibrium relationship and Granger causal relation between the observed number of tourists and a set of related keywords in the Baidu Index. It indicates a positive correlation between the increasing Baidu keyword search index and the increasing observed tourist flow. In our study, we first build a predication model based on a autoregressive moving average （ARMA） with baseline features of visitors＇ number. We then use a autoregressive distributed lag model（ARDL） by including the Baidu Index. The ARDL model improves the prediction accuracy of the training sample by 12.4% , and the testing sample by 14.5%. Our approach can predict the number of daily visitors of the Forbidden City using the one or two days lagging data from the Baidu Index, while the previous forecasting method requires data of a much longer period. In conclusion, it improves the timeliness and accuracy of the prediction, and provides tourism management departments with better evidence for decision-making. The governmentally published data can only reflect a few narrow aspects of the visitors＇ needs. The large volume of various unstructured data obtained from the Internet is more comprehensive and timely. The analytical model based on these data has better precision in tourist flow prediction. Some valuable information, such as actual desire and action of visitors, which are hardly presented by the structural data, can be extracted as well. To the best of our knowledge, this paper presents the first attempt to construct a model for correlating Internet search data based on the Baidu Index and actual tourism flow, and provides a new perspective for tourist flow prediction research.