A Multi-scale Combined Prediction on Tourist Flow: Does Internet Big Data Matter？

Abstract

Abstract This article proposes a multi-scale combination prediction method for tourism traffic driven by internet big data. Firstly, process data on tourist traffic, Baidu index, and online reviews. Use Principal Component Analysis (PCA) to reduce the dimensionality of Baidu Index, and then use Snownlp to calculate the emotional value of each online comment. Secondly, a mixed multi-scale decomposition method is used to decompose tourism traffic and processed data, and sample entropy is used to reconstruct the decomposition results into high-frequency, low-frequency, and trend components. Thirdly, LSTM, BPNN, and SVR are used to predict high-frequency, low-frequency, and trend components, respectively, to obtain the predicted values under the three decomposition paths. The optimal weighted combination is used to obtain the final prediction result. Finally, the empirical analysis of tourism flow in Jiuzhaigou Valley,the result shows that the Internet big data can significantly improve the prediction effect of tourism flow.