Data mining of tree-based models to analyze freeway accident frequency

Chang, Li-Yen; Chen, WenChieh

doi:10.1016/j.jsr.2005.06.013

Cited by 310 publications

(140 citation statements)

References 23 publications

Supporting

Mentioning

132

Contrasting

Unclassified

Order By: Relevance

“…The application of CART in road safety analysis has been advocated by many authors (Abdel--Aty et al, 2005;Council and Stewart, 1996;Chang and Chen, 2005;Chang and Wang, 2006;Chen and Jovanis, 2000;Kuhnert et al 2000;Magazzù et al 2006;Pande et al 2010;Qin and Han, 2008;Sohn and Shin 2001;Yan and Radwan, 2006;Yan et al 2010). The method has also been used for analysing other aspects of traffic engineering: used decision trees to forecast trip generation; considered using trees to determine modal correction factors for motor vehicle emissions; and Hallmark et al (2002) used trees to identify geometric and operational roadway characteristics that influenced vehicle activity.…”

Section: Introductionmentioning

confidence: 99%

A classification tree approach to identify key factors of transit service quality

Oña

Calvo

2012

Expert Systems with Applications

150

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

A classification tree approach to identify key factors of transit service quality

Oña

Calvo

2012

Expert Systems with Applications

150

View full text Add to dashboard Cite

“…In recent years, the tree-based algorithm has been gradually used in the analysis of transportation studies for analyzing the traffic accident rates and injury severity problems [43][44][45], investigating the travel behavior and transport mode choice [46,47] and identifying the interaction of different transportation characteristics [48].…”

Section: The Hierarchical Tree-based Regression Methods For Survey Datmentioning

confidence: 99%

The Impact of Beijing Subway’s New Fare Policy on Riders’ Attitude, Travel Pattern and Demand

Zhang

Yan

An³

et al. 2017

Sustainability

View full text Add to dashboard Cite

Abstract:On 28 December 2014, the Beijing subway's fare policy was changed from "Two Yuan" per trip to the era of Logging Ticket Price, charging users by travel mileage. This paper aims at investigating the effects of Beijing subway's new fare policy on the riders' attitude, travel pattern and demand. A survey analysis was conducted to identify the effects of the new fare policy for Beijing subway on riders' satisfaction degree and travel pattern associated with the potential influencing factors using Hierarchical Tree-based Regression (HTBR) models. The model results show that income, travel distance and month of travel have significant impacts on the subway riders' satisfaction degree, while trip purpose, car ownership and travel frequency significantly influence the riders' stated travel pattern. Overall, the degree of satisfaction could not be effectively recovered within five months after the new fare policy, but the negative public attitude did not depress the subway demand continuously. Based on the further time sequence analyses of the passenger flow volume data for two years, it is concluded that the new policy made the ridership decrease sharply in the first month but gradually came back to the previous level four months later, and then the passenger flow volume kept steady again. The findings in this study indicate that the new fare policy realized the purpose of lowering the government's financial pressure but did not reduce the subway ridership in a long term perspective.

show abstract

“…The tree size of producing the minimal cross validation evaluation method is determined to be the final model. More details about CART analysis and application can be found in Breiman et al [3], Chang and Chen [4].…”

Section: Methodology 31 Cart Methodsmentioning

confidence: 99%

“…Classification and Regression Tree (CART), one of the most popular data mining techniques, was introduced by Breiman et al [3], and has been applied in business administration, medicine, industry, and engineering fields [4]. CART is an interesting and effective non-parameter classification and regression method, in which the binary tree is established to recursively partition the data into smaller and smaller strata so as to improve the fit as best as possible.…”

Section: Introductionmentioning

confidence: 99%