基于机器学习和百度搜索指数的客流量预测  

Passenger Traffic Forecast Based on Machine Learning and Baidu Search Index

作　　者：李晓楠 史尚涵 施雪姣 Xiaonan Li;Shanghan Shi;Xuejiao Shi(School of Big Data Science,Hebei Finance University,Baoding Hebei)

机构地区：[1]河北金融学院大数据科学学院,河北保定

出　　处：《运筹与模糊学》2024年第6期963-973,共11页Operations Research and Fuzziology

摘　　要：随着互联网的迅速发展和旅游业的兴起,以百度搜索指数为基础的客流量预测可以帮助景区通过某些关键词的搜索量来判断游客的需求与关注度,及时做出改进与对策,从而高效而准确地服务游客。文章以预测九寨沟的日游客量为目标,通过初选、拓展,最后利用皮尔逊相关系数及时差相关法筛选出与九寨沟客流量紧密相关的关键词。接着将这些数据进行对数变换,利用随机森林、支持向量机回归、BP神经网络三种机器学习方法分别来预测九寨沟客流量。以决定系数R2和均方误差ME为评价指标并对比三个模型的结果值,择出最优预测模型。结果表明:BP神经网络的ME为0.019986,R2为0.83669,相较于另外两种预测模型而言,对九寨沟客流量的预测效果最好,可以帮助景区做好资源规划和应急准备。With the rapid development of the Internet and the rise of tourism,the passenger flow prediction based on Baidu search index index can help scenic spots to judge the needs and attention of tourists through the search volume of some keywords,and make timely improvements and countermeasures,so as to serve tourists efficiently and accurately.The article aims to predict the daily number of tourists in Jiuzhaigou,and through the primary selection and expansion,it finally selects the key words closely related to the passenger flow of Jiuzhaigou by means of Pearson’s correlation coefficient.Then the data were log-transformed,and three machine learning methods of random forest,support vector machine regression and BP neural network were used to predict the passenger flow in Jiuzhaigou respectively.Using the determination coefficient R2 and the mean squared error ME as the evaluation indexes and comparing the result values of the three models to select the optimal prediction models.The results show that the ME of BP neural network is 0.019986 and R2 is 0.83669.Compared with the other two prediction models,the prediction effect of Jiuzhaigou passenger flow is the best,which can help the scenic spot to make resource planning and emergency preparedness.

关 键 词：百度搜索指数 客流量预测 机器学习 时差相关法 

分 类 号：F59[经济管理—旅游管理]