The Analysis of Regional Ice and Snow Tourist Destinations under Back Propagation Neural Network

Abstract

This study aims to analyze the evolutionary characteristics and development levels of regional ice and snow tourist destinations by integrating the Back Propagation Neural Network (BPNN) within an Internet of Things (IoT) framework. Data from multiple sources are gathered through web scraping technology from various online platforms and are then subjected to cleaning, standardization, and normalization. A feature recognition model for ice and snow tourism is constructed based on a BPNN combined with a Spatial-Temporal Graph Convolutional Network (ST-GCN) algorithm. Experimental results demonstrate that this model excels in convergence speed and prediction accuracy, achieving a final convergence value of 0.059 and a prediction accuracy of 95.74%, which is at least 4% higher than that of the baseline BPNN algorithm. Additionally, the model yields Recall and F1 scores of 91.57% and 89.31%, respectively. After 98 iterations, the Root Mean Square Error (RMSE) is 6.26, significantly outperforming other model algorithms. These results indicate that the proposed model offers substantial advantages in enhancing the management and service quality of ice and snow tourist destinations, thereby providing valuable technical support and guidance for future intelligent tourism management.