Method for predicting tourist traffic attraction based on machine learning

In order to scientifically solve the nonlinear dynamic coupling and ecological constraint problems in tourist attractions’ passenger flow forecast, this study proposes a regularized extreme learning machine (IMPA-RELM) model based on improved marine predator algorithm optimization. A model performance comparison experiment was conducted with Qinghai Lake and Guilin Lijiang as research objects. The experiments showed that in the prediction of the Lijiang scenic area, MAE (1125.47) decreased by 18.1% compared with the baseline LSTM, and RMSE (1966.32) decreased by 12.7%; in the prediction of the ecological pulse passenger flow of Qinghai Lake, MAPE (6.811%) broke through the threshold bottleneck of the traditional model of 7%. The research results show that the passenger flow prediction method proposed in this paper has more obvious performance advantages than traditional baseline models.