Tourism demand forecasting using PCA-BPNN

Abstract

Accurate prediction of tourism demand is critically important for the efficient allocation of resources in scenic areas and managing sudden events. This paper presents a new tourism demand prediction model, PCA-BPNN neural network model. It utilizes Principal Component Analysis (PCA) to reduce the dimensionality of the collected Baidu Index data and mitigate overfitting issues. The model then constructs a backpropagation neural network (BPNN). Empirical research demonstrates that PCA-BPNN effectively identifies the nonlinear relationship between search keywords and the number of tourist arrivals and outperforms all benchmark models in terms of predictive performance.