基于LOF-SSA-PSO-LSSVM模型的旅游客流量预测  

Tourist flow prediction based on LOF-SSA-PSO-LSSVM model

作　　者：张娟 ZHANG Juan(College of Mathematics and Statistics,Hulunbuir College,021008,Hulunbuir,Inner Mongolia,PRC)

机构地区：[1]呼伦贝尔学院数学与统计学院,内蒙古自治区呼伦贝尔市021008

出　　处：《曲阜师范大学学报（自然科学版）》2023年第3期31-36,共6页Journal of Qufu Normal University(Natural Science)

基　　金：2022年呼伦贝尔学院科研资助项目(2022FDYB03);内蒙古自治区高等学校科学技术研究项目(NJZY19231)。

摘　　要：针对月度旅游客流量影响因素复杂,时间序列数据中可能存在异常值和噪声的特点,提出了一个基于局部异常因子(LOF)和奇异谱分析(SSA)的LOF-SSA-PSO-LSSVM预测模型.数据预处理阶段,对原始序列运用X12季节加法模型消除季节性的影响,采用LOF方法检测季节调整后序列的异常值,并使用线性插值和最小二乘支持向量机(LSSVM)方法来修正异常值,新的待预测序列由修正后的季节调整序列与季节因子序列加和得到.预测阶段,首先采用奇异谱分析(SSA)方法来剔除新的待预测序列中的噪声,然后采用PSO-LSSVM方法对其进行预测.以上海2004年1月至2018年12月旅游客流量序列的预测为例,通过与ARIMA、LSSVM等模型比较,表明LOF-SSA-PSO-LSSVM模型能有效提高旅游客流量的预测精度和稳定性.The analysis and prediction of tourist traffic is of great significance to the sustainable development of regional tourism.In view of the complexity of the influence factors of monthly tourist traffic and the characteristics of possible outliers and noise in the time series data,a LOF-SSA-PSO-LSSVM prediction model based on local anomaly factor(LOF)and singular spectral analysis(SSA)is proposed.In the data preprocessing stage,the original sequence is decomposed by X12 season addition model to eliminate seasonal effects,the outlier of the seasonally adjusted sequence is detected by LOF method,and the outlier is corrected by interpolation and least square support vector machine(LSSVM)method.The new predictable sequence is added by the revised seasonal adjustment sequence with the seasonal factor sequence.In the prediction phase,the singular spectral analysis(SSA)method is used to eliminate the noise in the new prediction sequence,and then the PSO-LSSVM method is used to predict it.Taking the forecast of the tourist traffic sequence from January 2004 to December 2018 in Shanghai as an example,it is shown that the LOF-SSA-PSO-LSSVM model can effectively improve the prediction accuracy and stability of tourist traffic by comparison with ARIMA and LSSVM models.

关 键 词：旅游客流量预测 局部异常因子 最小二乘支持向量机 粒子群寻优 奇异谱分析 

分 类 号：F592[经济管理—旅游管理] TP391[经济管理—产业经济]