Research On Tourist Destination Passenger Flow Prediction Based On Decomposition Clustering Technology For Data Continuity Problem

Tourist flow prediction faces challenges in achieving high accuracy due to the complex and dynamic nature of tourism data. This study addresses these challenges by proposing a multi-feature combined prediction model, Complete Ensemble Empirical Mode Decomposition with Adaptive Noise-Incremental Entropy-K-means clustering-Time Convolutional Network-Coati Optimization Algorithm (CEEMDAN-IEK-TT-COA). The model integrates signal decomposition, clustering reconstruction, component prediction, and optimization algorithms. Key features are selected through feature engineering, followed by data decomposition and reconstruction using CEEMDAN, Incremental Entropy, and K-means clustering. A model pool of Time Convolutional Network (TCN), Bidirectional Gated Recurrent Unit (Bi-GRU), Long Short-Term Memory network (LSTM), and Transformer filters optimal sub-models for combined prediction. The Coati Optimization Algorithm (COA) refines the final prediction. Experimental results on Jiuzhaigou and Siguniangshan datasets demonstrate significant RMSE improvements of $21.81 \%$ and $17.84 \%$, validating the model’s effectiveness.