Article: Study on accurate prediction method for daily tourist flow in tourist attractions based on feature recursive elimination Journal: International Journal of Computer Applications in Technology (IJCAT) 2026 Vol.78 No.6 pp.11 - 18 Abstract: To reduce Peak Lag Deviation (PLD), enhance Flow Mutation Responsiveness (FMR) and optimise Hotspot Overlap Rate (HOR), this paper proposes a feature recursive elimination-based method for accurate daily tourist flow prediction in attractions. Firstly, integrate historical data through data dimensionality reduction processing to reduce data complexity. Then, extract the daily average tourist volume and flow fluctuation characteristics and combine random forest and out of bag error estimation methods to recursively eliminate redundant features. Finally, using evolutionary strategies to optimise the weights and biases of the BP network model, the processed features are used as inputs to achieve accurate prediction through crossover and mutation operations. The experiment shows that for weekdays, the PLD, FMR and HOR of this method are 2.34%, 94.56% and 92.34%, respectively. For holidays, the PLD, FMR and HOR of this method are 5.62%, 83.21% and 81.23%, respectively. The numerical results are superior to existing methods. Inderscience Publishers - linking academia, business and industry through research

Title: Study on accurate prediction method for daily tourist flow in tourist attractions based on feature recursive elimination

Authors: Xiaoyan Wang

Addresses: Department of Economics and Trade, Yongcheng Vocational College, Yongcheng, Henan Province, China

Abstract: To reduce Peak Lag Deviation (PLD), enhance Flow Mutation Responsiveness (FMR) and optimise Hotspot Overlap Rate (HOR), this paper proposes a feature recursive elimination-based method for accurate daily tourist flow prediction in attractions. Firstly, integrate historical data through data dimensionality reduction processing to reduce data complexity. Then, extract the daily average tourist volume and flow fluctuation characteristics and combine random forest and out of bag error estimation methods to recursively eliminate redundant features. Finally, using evolutionary strategies to optimise the weights and biases of the BP network model, the processed features are used as inputs to achieve accurate prediction through crossover and mutation operations. The experiment shows that for weekdays, the PLD, FMR and HOR of this method are 2.34%, 94.56% and 92.34%, respectively. For holidays, the PLD, FMR and HOR of this method are 5.62%, 83.21% and 81.23%, respectively. The numerical results are superior to existing methods.

Keywords: tourist attractions; daily tourist flow; prediction methods; feature extraction; recursive elimination; evolutionary strategy; BP network.

DOI: 10.1504/IJCAT.2026.153739

International Journal of Computer Applications in Technology, 2026 Vol.78 No.6, pp.11 - 18

Received: 23 Jul 2025
Accepted: 14 Nov 2025
Published online: 22 May 2026 *

Title: Study on accurate prediction method for daily tourist flow in tourist attractions based on feature recursive elimination

Keep up-to-date