Data mining is the analysis of large 'observational' datasets to find unsuspected relationships that might be useful to the data owner. It typically involves analysis where objectives of the mining exercise have no bearing on the data collection strategy. Freeway traffic surveillance data collected through underground loop detectors is one such 'observational' database maintained for various ITS (Intelligent Transportation Systems) applications such as travel time prediction etc. In this research data mining process is used to relate this surrogate measure of traffic conditions with rear-end crash occurrence on freeways. Crash and dual loop detector data from 36.25 mile instrumented Interstate4 corridor in Orlando (FL) are used in this study. The research problem is set up as a classification problem and separate data mining based classifiers are developed to discriminate crashes belonging to different categories from normal conditions on the freeway. Based on the models developed in this study one can identify the traffic conditions prone to rear-end crashes 5-10 minutes prior to the crash. The findings of this research are proposed to be used as a proactive traffic management system which could warn the drivers about potential rear-end crashes. The paper presents a step by step approach of data analysis to develop a strategy to identify real-time traffic conditions prone to rear-end crashes using freeway loop detector data. It was concluded that the rear-end crashes on the freeway may be grouped into two distinct clusters based on the average speeds prevailing in approximately 2 mile section around the crash location 5-10 minutes before a crash. One cluster (group) of crashes occurs under extended congestion on the freeway while the average speeds are relatively higher during the 5-10 minute period before a cluster 2 crash (refer Table 1 for specific traffic speed conditions for each group of rear-end crashes). It was noticed that conditions belonging to cluster 1 occur very rarely and hence whenever such conditions are encountered in real-time then a crash warning may be issued. For cluster 2 rear-end crashes further neural network based classification models were developed. Based on the performance of the classification models and the proposed real-time application strategy, 75% of the rear-end crashes may be identified 5-10 minutes before their occurrence with just 34% positive decisions (i.e., crash warnings). Since crashes (however frequent on the 1-4 corridor under consideration) are rare events; these positive decisions would result in a significant number of 'false alarms'. However, it should be noted that 'false alarms' are not as detrimental in the present application as they are for incident detection algorithms. Crash prone traffic conditions, which have been identified in this paper, would not always result in a rear-end crash occurrence. The conditions, however, are worth warning the drivers and drivers need to be more attentive under such traffic conditions even if they may not always culminate in a rear-end crash. A reasonable number of warnings, which the drivers do not consider excessive, based on the models developed can potentially play a critical role in proactive traffic management. These warnings may be issued to the motorists driving on the freeway locations through VMS (variable message signs). However, the frequency and impacts of such warnings on driver behavior would need to be carefully estimated before implementing such measure. Another application for the findings of this research could be the formulations of VSL (variable speed limit) implementation strategies that can reduce the probability of rear-end crashes.
Application of data mining techniques for real-time crash risk assessment on freeways
2006
7 Seiten, 1 Bild, 1 Tabelle, 13 Quellen
Aufsatz (Konferenz)
Englisch
Application of Data Mining Techniques for Real-Time Crash Risk Assessment of Freeways
British Library Conference Proceedings | 2005
|Estimation of Real-Time Crash Risk: Are All Freeways Created Equal?
Online Contents | 2011
|Development of a crash risk index to identify real time crash risks on freeways
Online Contents | 2013
|