Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor Approach
Short-term prediction of passengers' flow is one of the essential elements of the operation and real time control for public transit. Although fine prediction methodologies have been reported, they still need improvement in terms of accuracy when the current or future data either exhibit fluctu...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2019-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/8811477/ |
id |
doaj-c0dd9a9b6e1d4e1f8d3aa35c8fe697ec |
---|---|
record_format |
Article |
spelling |
doaj-c0dd9a9b6e1d4e1f8d3aa35c8fe697ec2021-03-30T00:03:07ZengIEEEIEEE Access2169-35362019-01-01712093712094910.1109/ACCESS.2019.29371148811477Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor ApproachShidong Liang0https://orcid.org/0000-0002-2191-6187Minghui Ma1https://orcid.org/0000-0002-7080-4376Shengxue He2Hu Zhang3Business School, University of Shanghai for Science and Technology, Shanghai, ChinaSchool of Mechanical and Automotive Engineering, Shanghai University of Engineering Science, Shanghai, ChinaBusiness School, University of Shanghai for Science and Technology, Shanghai, ChinaBusiness School, University of Shanghai for Science and Technology, Shanghai, ChinaShort-term prediction of passengers' flow is one of the essential elements of the operation and real time control for public transit. Although fine prediction methodologies have been reported, they still need improvement in terms of accuracy when the current or future data either exhibit fluctuations or significant change. To address this issue, in this study, a fusion method including Kalman filtering and K-Nearest Neighbor approach is proposed. The core point of this method is to design a framework to dynamically adjust the weight coefficients of the predicted values obtained by Kalman filtering and K-Nearest Neighbor approach. The Kalman filtering and K-Nearest Neighbor approach can handle different variation trend of the data. The dynamic weight coefficient can more accurately predict the final value by giving more weight to the appropriately predicted method. In the case study of real-world data, the predicted values of alighting passengers and boarding passengers are presented by four predicted methods involving Kalman filtering, K-Nearest Neighbor approach, support vector machine, and the proposed method. According to the comparison of the test results, the proposed fusion method performed better in terms of predicting accuracy, even if time-series data abruptly varied or exhibited wide fluctuations. The proposed methodology was found as one of the effective approaches based on the historical data and current data in the area of passengers' flow forecasting for urban public transit.https://ieeexplore.ieee.org/document/8811477/Short-term forecastingurban public transitpassenger flowfusion model |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Shidong Liang Minghui Ma Shengxue He Hu Zhang |
spellingShingle |
Shidong Liang Minghui Ma Shengxue He Hu Zhang Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor Approach IEEE Access Short-term forecasting urban public transit passenger flow fusion model |
author_facet |
Shidong Liang Minghui Ma Shengxue He Hu Zhang |
author_sort |
Shidong Liang |
title |
Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor Approach |
title_short |
Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor Approach |
title_full |
Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor Approach |
title_fullStr |
Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor Approach |
title_full_unstemmed |
Short-Term Passenger Flow Prediction in Urban Public Transport: Kalman Filtering Combined K-Nearest Neighbor Approach |
title_sort |
short-term passenger flow prediction in urban public transport: kalman filtering combined k-nearest neighbor approach |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2019-01-01 |
description |
Short-term prediction of passengers' flow is one of the essential elements of the operation and real time control for public transit. Although fine prediction methodologies have been reported, they still need improvement in terms of accuracy when the current or future data either exhibit fluctuations or significant change. To address this issue, in this study, a fusion method including Kalman filtering and K-Nearest Neighbor approach is proposed. The core point of this method is to design a framework to dynamically adjust the weight coefficients of the predicted values obtained by Kalman filtering and K-Nearest Neighbor approach. The Kalman filtering and K-Nearest Neighbor approach can handle different variation trend of the data. The dynamic weight coefficient can more accurately predict the final value by giving more weight to the appropriately predicted method. In the case study of real-world data, the predicted values of alighting passengers and boarding passengers are presented by four predicted methods involving Kalman filtering, K-Nearest Neighbor approach, support vector machine, and the proposed method. According to the comparison of the test results, the proposed fusion method performed better in terms of predicting accuracy, even if time-series data abruptly varied or exhibited wide fluctuations. The proposed methodology was found as one of the effective approaches based on the historical data and current data in the area of passengers' flow forecasting for urban public transit. |
topic |
Short-term forecasting urban public transit passenger flow fusion model |
url |
https://ieeexplore.ieee.org/document/8811477/ |
work_keys_str_mv |
AT shidongliang shorttermpassengerflowpredictioninurbanpublictransportkalmanfilteringcombinedknearestneighborapproach AT minghuima shorttermpassengerflowpredictioninurbanpublictransportkalmanfilteringcombinedknearestneighborapproach AT shengxuehe shorttermpassengerflowpredictioninurbanpublictransportkalmanfilteringcombinedknearestneighborapproach AT huzhang shorttermpassengerflowpredictioninurbanpublictransportkalmanfilteringcombinedknearestneighborapproach |
_version_ |
1724188699961327616 |