Dynamic Traffic Demand Forecasting Based on Microzone: A Spatial Analysis of Smart Card Data Using Machine Learning Methodologies.
Saved in:
| Title: | Dynamic Traffic Demand Forecasting Based on Microzone: A Spatial Analysis of Smart Card Data Using Machine Learning Methodologies. |
|---|---|
| Authors: | Kim, Kyoungtae1 (AUTHOR), Lee, Inmook2 (AUTHOR), Choi, Solsaem2 (AUTHOR), Song, Ki-Han2 (AUTHOR) kihansong@seoultech.ac.kr, Habib, Mohammad Rezwan (AUTHOR) mohabib@wiley.com |
| Source: | Journal of Advanced Transportation. 6/23/2026, Vol. 2026, p1-18. 18p. |
| Subjects: | Machine learning, Geographic spatial analysis, Public transit, Spatial analysis (Statistics), Traffic estimation, Random forest algorithms, Bayesian analysis, Smart cards |
| Abstract: | This study seeks to develop an innovative dynamic model for predicting public transportation traffic volume. Current static traffic demand forecasting models suffer from significant drawbacks, as they struggle to adapt to real‐time changes due to their rigid reliance on fixed time intervals and limited datasets. To overcome these constraints, the research strategically divides regions into microzones and constructs a flexible forecasting model by examining traffic volume variations across microzones and trip contexts, leveraging smart card data and comprehensive socioeconomic indicators. To improve the precision of forecasting, a SHapley Additive ExPlanations (SHAP) analysis was performed to pinpoint the most significant variables. These key variables were subsequently re‐extracted from the original dataset and utilized as input parameters for the model. The primary independent variables encompassed traffic‐related factors, including transfer frequency, direct distance between the origin and destination points, and integrated stop IDs for both the starting and ending locations. Furthermore, the analysis incorporated socioeconomic indicators such as population size, household count, workforce numbers, sales area, associative area, and the number of hospitals at both the origin and destination points. To uncover the spatial dynamics of traffic volume, a random forest model was utilized to evaluate the significance of critical variables and predict traffic flow patterns. Additionally, Bayesian linear regression and ordinary least squares (OLS) analyses were implemented to examine explanatory relationships and explore the nuanced characteristics of direct demand models. The predictive capabilities of each forecasting approach were subsequently compared and assessed. In this study, dynamic refers to an operationally updatable microzone forecasting framework rather than an explicit time‐step time‐series model. This research aims to address urban mobility challenges by developing a sophisticated approach to predict traffic volumes. By conducting detailed spatial analysis, identifying critical influencing factors, and creating precise forecasting models at the microzone level, the study seeks to help reduce traffic congestion and optimize public transportation system performance in real time. [ABSTRACT FROM AUTHOR] |
| Copyright of Journal of Advanced Transportation is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | This study seeks to develop an innovative dynamic model for predicting public transportation traffic volume. Current static traffic demand forecasting models suffer from significant drawbacks, as they struggle to adapt to real‐time changes due to their rigid reliance on fixed time intervals and limited datasets. To overcome these constraints, the research strategically divides regions into microzones and constructs a flexible forecasting model by examining traffic volume variations across microzones and trip contexts, leveraging smart card data and comprehensive socioeconomic indicators. To improve the precision of forecasting, a SHapley Additive ExPlanations (SHAP) analysis was performed to pinpoint the most significant variables. These key variables were subsequently re‐extracted from the original dataset and utilized as input parameters for the model. The primary independent variables encompassed traffic‐related factors, including transfer frequency, direct distance between the origin and destination points, and integrated stop IDs for both the starting and ending locations. Furthermore, the analysis incorporated socioeconomic indicators such as population size, household count, workforce numbers, sales area, associative area, and the number of hospitals at both the origin and destination points. To uncover the spatial dynamics of traffic volume, a random forest model was utilized to evaluate the significance of critical variables and predict traffic flow patterns. Additionally, Bayesian linear regression and ordinary least squares (OLS) analyses were implemented to examine explanatory relationships and explore the nuanced characteristics of direct demand models. The predictive capabilities of each forecasting approach were subsequently compared and assessed. In this study, dynamic refers to an operationally updatable microzone forecasting framework rather than an explicit time‐step time‐series model. This research aims to address urban mobility challenges by developing a sophisticated approach to predict traffic volumes. By conducting detailed spatial analysis, identifying critical influencing factors, and creating precise forecasting models at the microzone level, the study seeks to help reduce traffic congestion and optimize public transportation system performance in real time. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 01976729 |
| DOI: | 10.1155/atr/5849967 |
