The University of Southampton
University of Southampton Institutional Repository

A random forest incident detection algorithm that incorporates contexts

A random forest incident detection algorithm that incorporates contexts
A random forest incident detection algorithm that incorporates contexts
A major problem faced by state of the art incident detection algorithms is their high false alert rates, which are caused in part by failing to differentiate incidents from contexts. Contexts are referred to as external factors that could be expected to influence traffic conditions, such as sporting events, public holidays and weather conditions. This paper presents RoadCast Incident Detection (RCID), an algorithm that aims to make this differentiation by gaining a better understanding of conditions that could be expected during contexts’ disruption. RCID is based on a previously developed random forest traffic forecasting algorithm, RoadCast, which uses contextual data to create forecasts of traffic conditions that could be expected if no incident occurred. RCID compares these forecasts with real-time conditions, and raises alerts when there is a sufficient difference. RCID was evaluated on loop detector flow data and city council incident logs from Southampton, U.K. Comparisons were made with and without context, and to a state of the art algorithm, RAID. RCID was found to outperform RAID in terms of detection rate and false alert rate. RCID was also found to have a 25% lower false alert rate when incorporating contextual data. This improvement suggests that if RCID were to be implemented in a Traffic Management Centre, operators would be distracted by far fewer false alerts from contexts than is currently the case with state of the art algorithms, and so could detect incidents more effectively.
Random forest, Traffic flow prediction, Big data, Machine Learning, Context
Evans, Jonny, Rhys Alexander
fa85f2a7-dd18-4b43-9b40-d0f40bb7363d
Waterson, Ben
60a59616-54f7-4c31-920d-975583953286
Hamilton, Andrew
12ead9ac-0af5-4773-a657-906b4d89772b
Evans, Jonny, Rhys Alexander
fa85f2a7-dd18-4b43-9b40-d0f40bb7363d
Waterson, Ben
60a59616-54f7-4c31-920d-975583953286
Hamilton, Andrew
12ead9ac-0af5-4773-a657-906b4d89772b

Evans, Jonny, Rhys Alexander, Waterson, Ben and Hamilton, Andrew (2019) A random forest incident detection algorithm that incorporates contexts. 15th World Conference on Transport Research, Mumbai, India. 26 - 31 May 2019. 14 pp .

Record type: Conference or Workshop Item (Paper)

Abstract

A major problem faced by state of the art incident detection algorithms is their high false alert rates, which are caused in part by failing to differentiate incidents from contexts. Contexts are referred to as external factors that could be expected to influence traffic conditions, such as sporting events, public holidays and weather conditions. This paper presents RoadCast Incident Detection (RCID), an algorithm that aims to make this differentiation by gaining a better understanding of conditions that could be expected during contexts’ disruption. RCID is based on a previously developed random forest traffic forecasting algorithm, RoadCast, which uses contextual data to create forecasts of traffic conditions that could be expected if no incident occurred. RCID compares these forecasts with real-time conditions, and raises alerts when there is a sufficient difference. RCID was evaluated on loop detector flow data and city council incident logs from Southampton, U.K. Comparisons were made with and without context, and to a state of the art algorithm, RAID. RCID was found to outperform RAID in terms of detection rate and false alert rate. RCID was also found to have a 25% lower false alert rate when incorporating contextual data. This improvement suggests that if RCID were to be implemented in a Traffic Management Centre, operators would be distracted by far fewer false alerts from contexts than is currently the case with state of the art algorithms, and so could detect incidents more effectively.

Full text not available from this repository.

More information

Accepted/In Press date: 22 November 2018
Published date: 26 May 2019
Venue - Dates: 15th World Conference on Transport Research, Mumbai, India, 2019-05-26 - 2019-05-31
Keywords: Random forest, Traffic flow prediction, Big data, Machine Learning, Context

Identifiers

Local EPrints ID: 426475
URI: https://eprints.soton.ac.uk/id/eprint/426475
PURE UUID: 08668d78-6026-4ae1-96b0-78edc9c5f5f7
ORCID for Ben Waterson: ORCID iD orcid.org/0000-0001-9817-7119

Catalogue record

Date deposited: 28 Nov 2018 17:30
Last modified: 10 Sep 2019 00:50

Export record

Download statistics

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics

Atom RSS 1.0 RSS 2.0

Contact ePrints Soton: eprints@soton.ac.uk

ePrints Soton supports OAI 2.0 with a base URL of https://eprints.soton.ac.uk/cgi/oai2

This repository has been built using EPrints software, developed at the University of Southampton, but available to everyone to use.

We use cookies to ensure that we give you the best experience on our website. If you continue without changing your settings, we will assume that you are happy to receive cookies on the University of Southampton website.

×