摘要
利用网络搜索数据进行短期旅游客流量预测,对景区优化资源的调度、引导旅游地规划开发和指导游客制定出行计划都具有重要意义。针对噪声对预测效果的干扰,将经验模态分解(EMD)去高频噪后的网络搜索数据作为解释变量,引入收敛速度快、训练时间短、算力消耗较小的门控循环单元(GRU)神经网络构建旅游客流量预测模型。以重庆黑山谷景区为例,选用2015年1月1日至2020年1月24日的网络搜索数据和游客数据进行实证分析。预测结果表明,相较于循环神经网络和长短期记忆模型,EMD-GRU模型的预测效能更高,且使用EMD去噪数据训练的预测模型能有效提高原始数据预测模型的精度。
Predicting the short-term tourists flow by the network search data is of great significance to optimize the resource scheduling of scenic spots,guide the planning and development of tourist destinations and guide tourists to make travel plans.In view of the interference of noise on the prediction effect,the network search data taken as the explanatory variable is denoised by Empirical Mode Decomposition(EMD).Gate Recurrent Unit(GRU)with fast convergence speed,short training time and low computational power consumption is introduced to construct the tourists flow prediction model.Taking Black Valley scenic area of Chongqing as an example,the network search data and tourists data from January 1,2015 to January 24,2020 are selected for empirical analysis.The prediction results show that EMD-GRU model has higher prediction efficiency when compared with Recurrent Neural Network(RNN)and Long Short Time Memory(LSTM),and the prediction model trained with EMD denoising data performs better in effectively improving the accuracy of original data prediction model.
作者
崔洪瑞
杨晓霞
余阳立
CUI Hong-rui;YANG Xiao-xia;YU Yang-li(School of Geographical Sciences,Southwest University,Chongqing 400715,China;Tourism Research Institute,Southwest University,Chongqing 400715,China)
出处
《西华师范大学学报（自然科学版）》
2023年第2期179-185,共7页
Journal of China West Normal University(Natural Sciences)
基金
重庆市社科规划一般项目(2017YBGL162)
西南大学西南山地生态循环农业国家级培育基地项目(5330200076)。
关键词
客流量预测
网络搜索
门控循环单元
经验模态分解
神经网络
重庆黑山谷
tourists flow prediction
network search
Gated Recurrent Unit(GRU)
Empirical Mode Decomposition(EMD)
neural network
Black Valley of Chongqing
