基于滤噪百度指数的北京客流量预测研究

Research on Beijing Tourism Forecasting Based on the Baidu Index with Noise Reduction

在搜索数据用于旅游客流量预测的研究中,对于搜索数据中的噪声干扰问题一直没有得到有效解决。以北京市旅游客流量预测为例,对比百度指数的预测能力和经过滤噪后的百度搜索预测能力,并以时间序列模型和BP神经网络模型作为基准模型参照。结果发现经过滤噪后的百度搜索预测能力和预测精度显著提高;研究发现,搜索数据在旅游客流量预测中,本地搜索的干扰和高峰期搜索特征对客流量预测结果都有较大的影响。

The research on tourism forecasting with search engines is relatively mature,but the noise interference in search data has not been effectively solved.Taking Beijing tourism forecast as an example,this paper compares the predictive ability of Baidu Index with and without filtering noise,and utilizes the time series model and BP neural network model as the benchmark for reference.As a result,it is found that the predictive ability and the prediction accuracy of Baidu Index after filtering the noise are significantly improved.In addition,we conclude that the interference of local search and the characteristics of peak search have a great impact in the forecasting of tourist flow.