A parking occupancy prediction method incorporating time series decomposition and temporal pattern attention mechanism

Parking occupancy prediction is an important reference for travel decisions and parking management. However, due to various related factors, such as commuting or traffic accidents, parking occupancy has complex change features that are difficult to model accurately, thus making it difficult for parking occupancy to be accurately predicted. Moreover, how to give appropriate weights to these changing features in prediction becomes a new challenge in the era of machine learning. To tackle these challenges, a parking occupancy prediction method called time series decomposition–long and short‐term memory neural network (LSTM)–temporal pattern attention mechanism, which consists of three modules, namely 1) time series decomposition: modelling parking occupancy changes by extracting features such as trend, period, and effect; 2) encoder: extracting temporal correlations of feature sequences with LSTM; 3) temporal pattern attention mechanism: assigning attention to different features, are proposed. The evaluation results of 30 parking lots in Guangzhou city show that the proposed model 1) improves accuracy over the baseline model LSTM by 9.14% on average; 2) performs outstanding in four prediction time intervals and six types of parking lots, proving its validity and generality; 3) demonstrates its rationality and interpretability through ablation experiments and Shapley additive explanation. We propose a parking occupancy prediction method based on the integration of time series decomposition (TSD), long and short‐term memory neural network (LSTM), and temporal pattern attention mechanism (TPA) . Specifically, the adoption of the TSD module enables the LSTM network to extract the temporal features more efficiently. Correspondingly, the TPA module gives appropriate weight to each temporal feature. We tested the accuracy, generalizability, and interpretability of the proposed method on a real‐world shared dataset with 30 parking lots in Guangzhou, China. The results show that TSD‐LSTM‐TPA achieves outstanding performance in four prediction time intervals and six types of parking lots. It is found that the trend and cycle features play a major role in prediction and quantified the importance of features in each input time step through ablation experiment and shapley additive explanations, respectively.