Context-aware online spatiotemporal traffic prediction

Jie Xu, Dingxiong Deng, Ugur Demiryurek, Cyrus Shahabi, Mihaela Van Der Schaar

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

With the availability of traffic sensors data, various techniques have been proposed to make congestion prediction by utilizing those datasets. One key challenge in predicting traffic congestion is how much to rely on the historical data v.s. The real-time data. To better utilize both the historical and real-time data, in this paper we propose a novel online framework that could learn the current situation from the real-time data and predict the future using the most effective predictor in this situation from a set of predictors that are trained using historical data. In particular, the proposed framework uses a set of base predictors (e.g. A Support Vector Machine or a Bayes classifier) and learns in real-time the most effective one to use in different contexts (e.g. Time, location, weather condition). As real-time traffic data arrives, the context space is adaptively partitioned in order to efficiently estimate the effectiveness of each predictor in different contexts. We obtain and prove both short-term and long-term performance guarantees (bounds) for our online algorithm. Our experiments with real-world data in real-life conditions show that the proposed approach significantly outperforms existing solutions.

Original languageEnglish (US)
Title of host publicationIEEE International Conference on Data Mining Workshops, ICDMW
PublisherIEEE Computer Society
Pages43-46
Number of pages4
Volume2015-January
EditionJanuary
DOIs
StatePublished - Jan 26 2015
Externally publishedYes
Event14th IEEE International Conference on Data Mining Workshops, ICDMW 2014 - Shenzhen, China
Duration: Dec 14 2014 → …

Other

Other14th IEEE International Conference on Data Mining Workshops, ICDMW 2014
CountryChina
CityShenzhen
Period12/14/14 → …

Fingerprint

Traffic congestion
Support vector machines
Classifiers
Availability
Sensors
Experiments

ASJC Scopus subject areas

  • Computer Science Applications
  • Software

Cite this

Xu, J., Deng, D., Demiryurek, U., Shahabi, C., & Schaar, M. V. D. (2015). Context-aware online spatiotemporal traffic prediction. In IEEE International Conference on Data Mining Workshops, ICDMW (January ed., Vol. 2015-January, pp. 43-46). [7022576] IEEE Computer Society. https://doi.org/10.1109/ICDMW.2014.102

Context-aware online spatiotemporal traffic prediction. / Xu, Jie; Deng, Dingxiong; Demiryurek, Ugur; Shahabi, Cyrus; Schaar, Mihaela Van Der.

IEEE International Conference on Data Mining Workshops, ICDMW. Vol. 2015-January January. ed. IEEE Computer Society, 2015. p. 43-46 7022576.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Xu, J, Deng, D, Demiryurek, U, Shahabi, C & Schaar, MVD 2015, Context-aware online spatiotemporal traffic prediction. in IEEE International Conference on Data Mining Workshops, ICDMW. January edn, vol. 2015-January, 7022576, IEEE Computer Society, pp. 43-46, 14th IEEE International Conference on Data Mining Workshops, ICDMW 2014, Shenzhen, China, 12/14/14. https://doi.org/10.1109/ICDMW.2014.102
Xu J, Deng D, Demiryurek U, Shahabi C, Schaar MVD. Context-aware online spatiotemporal traffic prediction. In IEEE International Conference on Data Mining Workshops, ICDMW. January ed. Vol. 2015-January. IEEE Computer Society. 2015. p. 43-46. 7022576 https://doi.org/10.1109/ICDMW.2014.102
Xu, Jie ; Deng, Dingxiong ; Demiryurek, Ugur ; Shahabi, Cyrus ; Schaar, Mihaela Van Der. / Context-aware online spatiotemporal traffic prediction. IEEE International Conference on Data Mining Workshops, ICDMW. Vol. 2015-January January. ed. IEEE Computer Society, 2015. pp. 43-46
@inproceedings{5a04949cac7840b691e29c95f352a660,
title = "Context-aware online spatiotemporal traffic prediction",
abstract = "With the availability of traffic sensors data, various techniques have been proposed to make congestion prediction by utilizing those datasets. One key challenge in predicting traffic congestion is how much to rely on the historical data v.s. The real-time data. To better utilize both the historical and real-time data, in this paper we propose a novel online framework that could learn the current situation from the real-time data and predict the future using the most effective predictor in this situation from a set of predictors that are trained using historical data. In particular, the proposed framework uses a set of base predictors (e.g. A Support Vector Machine or a Bayes classifier) and learns in real-time the most effective one to use in different contexts (e.g. Time, location, weather condition). As real-time traffic data arrives, the context space is adaptively partitioned in order to efficiently estimate the effectiveness of each predictor in different contexts. We obtain and prove both short-term and long-term performance guarantees (bounds) for our online algorithm. Our experiments with real-world data in real-life conditions show that the proposed approach significantly outperforms existing solutions.",
author = "Jie Xu and Dingxiong Deng and Ugur Demiryurek and Cyrus Shahabi and Schaar, {Mihaela Van Der}",
year = "2015",
month = "1",
day = "26",
doi = "10.1109/ICDMW.2014.102",
language = "English (US)",
volume = "2015-January",
pages = "43--46",
booktitle = "IEEE International Conference on Data Mining Workshops, ICDMW",
publisher = "IEEE Computer Society",
edition = "January",

}

TY - GEN

T1 - Context-aware online spatiotemporal traffic prediction

AU - Xu, Jie

AU - Deng, Dingxiong

AU - Demiryurek, Ugur

AU - Shahabi, Cyrus

AU - Schaar, Mihaela Van Der

PY - 2015/1/26

Y1 - 2015/1/26

N2 - With the availability of traffic sensors data, various techniques have been proposed to make congestion prediction by utilizing those datasets. One key challenge in predicting traffic congestion is how much to rely on the historical data v.s. The real-time data. To better utilize both the historical and real-time data, in this paper we propose a novel online framework that could learn the current situation from the real-time data and predict the future using the most effective predictor in this situation from a set of predictors that are trained using historical data. In particular, the proposed framework uses a set of base predictors (e.g. A Support Vector Machine or a Bayes classifier) and learns in real-time the most effective one to use in different contexts (e.g. Time, location, weather condition). As real-time traffic data arrives, the context space is adaptively partitioned in order to efficiently estimate the effectiveness of each predictor in different contexts. We obtain and prove both short-term and long-term performance guarantees (bounds) for our online algorithm. Our experiments with real-world data in real-life conditions show that the proposed approach significantly outperforms existing solutions.

AB - With the availability of traffic sensors data, various techniques have been proposed to make congestion prediction by utilizing those datasets. One key challenge in predicting traffic congestion is how much to rely on the historical data v.s. The real-time data. To better utilize both the historical and real-time data, in this paper we propose a novel online framework that could learn the current situation from the real-time data and predict the future using the most effective predictor in this situation from a set of predictors that are trained using historical data. In particular, the proposed framework uses a set of base predictors (e.g. A Support Vector Machine or a Bayes classifier) and learns in real-time the most effective one to use in different contexts (e.g. Time, location, weather condition). As real-time traffic data arrives, the context space is adaptively partitioned in order to efficiently estimate the effectiveness of each predictor in different contexts. We obtain and prove both short-term and long-term performance guarantees (bounds) for our online algorithm. Our experiments with real-world data in real-life conditions show that the proposed approach significantly outperforms existing solutions.

UR - http://www.scopus.com/inward/record.url?scp=84936880788&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84936880788&partnerID=8YFLogxK

U2 - 10.1109/ICDMW.2014.102

DO - 10.1109/ICDMW.2014.102

M3 - Conference contribution

VL - 2015-January

SP - 43

EP - 46

BT - IEEE International Conference on Data Mining Workshops, ICDMW

PB - IEEE Computer Society

ER -