Urban Parking Demand Forecasting Using xLSTM-Informer Model
With the rapid advancement of urbanization, traffic congestion and parking shortages have emerged as critical challenges for modern cities. Accurate parking demand forecasting plays a pivotal role in supporting intelligent traffic management, resource allocation, and the efficient operation of parki...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10982176/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850277052716941312 |
|---|---|
| author | Jimei Ma Zhong Wang |
| author_facet | Jimei Ma Zhong Wang |
| author_sort | Jimei Ma |
| collection | DOAJ |
| description | With the rapid advancement of urbanization, traffic congestion and parking shortages have emerged as critical challenges for modern cities. Accurate parking demand forecasting plays a pivotal role in supporting intelligent traffic management, resource allocation, and the efficient operation of parking facilities. While deep learning techniques have notably enhanced prediction capabilities, current studies often fail to effectively combine the strengths of diverse model architectures. In this study, we propose a hybrid forecasting model—xLSTM-Informer—which integrates the Extended Long Short-Term Memory Network (xLSTM) and the Informer architecture. To validate the model’s generalizability and scalability, we extend our analysis beyond a single commercial parking lot and include four additional sites representing diverse urban land-use types—office, residential, hospital, and transit hub areas—across different spatial scales. Extensive experiments conducted on real-world hourly parking datasets demonstrate that the proposed xLSTM-Informer model consistently outperforms baseline models. Notably, it achieves a mean absolute percentage error (MAPE) of 3.93% for vehicle entry and 8.66% for vehicle exit prediction, along with strong performance in R2, RMSE, and MAE metrics. These results highlight the model’s robustness, high accuracy, and potential for practical deployment in city-scale parking demand forecasting. |
| format | Article |
| id | doaj-art-dab2219fd4014e2d821c8265fefd6ab7 |
| institution | OA Journals |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-dab2219fd4014e2d821c8265fefd6ab72025-08-20T01:50:00ZengIEEEIEEE Access2169-35362025-01-0113806018061110.1109/ACCESS.2025.356645810982176Urban Parking Demand Forecasting Using xLSTM-Informer ModelJimei Ma0https://orcid.org/0009-0004-2846-3354Zhong Wang1Dalian University of Technology, Dalian, ChinaDalian University of Technology, Dalian, ChinaWith the rapid advancement of urbanization, traffic congestion and parking shortages have emerged as critical challenges for modern cities. Accurate parking demand forecasting plays a pivotal role in supporting intelligent traffic management, resource allocation, and the efficient operation of parking facilities. While deep learning techniques have notably enhanced prediction capabilities, current studies often fail to effectively combine the strengths of diverse model architectures. In this study, we propose a hybrid forecasting model—xLSTM-Informer—which integrates the Extended Long Short-Term Memory Network (xLSTM) and the Informer architecture. To validate the model’s generalizability and scalability, we extend our analysis beyond a single commercial parking lot and include four additional sites representing diverse urban land-use types—office, residential, hospital, and transit hub areas—across different spatial scales. Extensive experiments conducted on real-world hourly parking datasets demonstrate that the proposed xLSTM-Informer model consistently outperforms baseline models. Notably, it achieves a mean absolute percentage error (MAPE) of 3.93% for vehicle entry and 8.66% for vehicle exit prediction, along with strong performance in R2, RMSE, and MAE metrics. These results highlight the model’s robustness, high accuracy, and potential for practical deployment in city-scale parking demand forecasting.https://ieeexplore.ieee.org/document/10982176/xLSTMinformerurban parking demand forecastingdeep learning |
| spellingShingle | Jimei Ma Zhong Wang Urban Parking Demand Forecasting Using xLSTM-Informer Model IEEE Access xLSTM informer urban parking demand forecasting deep learning |
| title | Urban Parking Demand Forecasting Using xLSTM-Informer Model |
| title_full | Urban Parking Demand Forecasting Using xLSTM-Informer Model |
| title_fullStr | Urban Parking Demand Forecasting Using xLSTM-Informer Model |
| title_full_unstemmed | Urban Parking Demand Forecasting Using xLSTM-Informer Model |
| title_short | Urban Parking Demand Forecasting Using xLSTM-Informer Model |
| title_sort | urban parking demand forecasting using xlstm informer model |
| topic | xLSTM informer urban parking demand forecasting deep learning |
| url | https://ieeexplore.ieee.org/document/10982176/ |
| work_keys_str_mv | AT jimeima urbanparkingdemandforecastingusingxlstminformermodel AT zhongwang urbanparkingdemandforecastingusingxlstminformermodel |