An Innovative Adaptive Pedestrian Flow Prediction Model Based on HMM-Fod-LSTM

This study introduces an innovative error-adjusted pedestrian flow prediction model, termed Fod-LSTM, which effectively addresses two key challenges in data prediction: data inconsistency and the large proportion of zero data, both of which can significantly compromise prediction accuracy. By seamlessly integrating the state probability statistical strategy from the Hidden Marker Model (HMM), the study constructs a robust hybrid model, HMM-Fod-LSTM. Experiment results show that the proposed model is superior to the other four hybrid models in terms of accuracy and crowding point accuracy. This robust performance underscores the model’s ability to handle the complexities and uncertainties inherent in the pedestrian flow prediction problem. This advancement holds promise for urban managers, enabling more informed decisions to alleviate congestion and improve residents’ quality of life.