ScholarLens Back to search
2017•Journal of the Transportation Research ForumRequires access

An Application of Decision Tree Models toExamine Motor Vehicle Crash Severity Outcomes

Jill M. Bernard

Open publisher page 2 citations

Abstract

Classification and Regression Tree (CART) and chi-square automatic interaction detection (CHAID) decision tree models are estimated and compared to examine the effect of driver characteristics and behaviors, temporal factors, weather conditions, and road characteristics on motor vehicle crash severity levels using Missouri crash data from 2002 to 2012. The CHAID model is found to significantly better discriminate among severity outcomes, and results suggest that the presence of alcohol, speeding, and failing to yield lead to many fatalities each year and likely have interactive effects. Decision rules are used to identify changes in driving policies expected to reduce severity outcomes.

About this research paper

What this paper is about

Classification and Regression Tree (CART) and chi-square automatic interaction detection (CHAID) decision tree models are estimated and compared to examine the effect of driver characteristics and behaviors, temporal factors, weather conditions, and road characteristics on motor vehicle crash severity levels using Missouri crash data from 2002 to 2012. The CHAID model is found to significantly better discriminate among severity outcomes, and results suggest that the presence of alcohol, speeding, and failing to yield lead to many fatalities each year and likely have interactive effects. Decision rules are used to identify changes in driving policies expected to reduce severity outcomes.

Why it matters

OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Classification and Regression Tree (CART) and chi-square automatic interaction detection (CHAID) decision tree models are estimated and compared to examine the effect of driver characteristics and behaviors, temporal factors, weather conditions, and road characteristics on motor vehicle crash severity levels using Missouri crash data from 2002 to 2012. The CHAID model is found to significantly better discriminate among severity outcomes, and results suggest that the presence of alcohol, speeding, and failing to yield lead to many fatalities each year and likely have interactive effects. Decision rules are used to identify changes in driving policies expected to reduce severity outcomes.

Key concepts: CHAID, Decision tree, Crash, Motor vehicle crash, Cart, Poison control, Decision tree learning, Predictive modelling

Related papers

Related paper 1Related paper 2Related paper 3
Back to paper searchBrowse research topicsOriginal source
An Application of Decision Tree Models toExamine Motor Vehicle Crash Severity Outcomes — Research Paper | ScholarLens