Griselda López

Analysis of traffic accidents on rural highways using Latent Class Clustering and Bayesian Networks

One of the principal objectives of traffic accident analyses is to identify key factors that affect the severity of an accident. However, with the presence of heterogeneity in the raw data used, the analysis of traffic accidents becomes difficult. In this paper, Latent Class Cluster (LCC) is used as a preliminary tool for segmentation of 3229 accidents on rural highways in Granada (Spain) between 2005 and 2008. Next, Bayesian Networks (BNs) are used to identify the main factors involved in accident severity for both, the entire database (EDB) and the clusters previously obtained by LCC. The results of these cluster-based analyses are compared with the results of a full-data analysis. The results show that the combined use of both techniques is very interesting as it reveals further information that would not have been obtained without prior segmentation of the data. BN inference is used to obtain the variables that best identify accidents with killed or seriously injured. Accident type and sight distance have been identify in all the cases analysed; other variables such as time, occupant involved or age are identified in EDB and only in one cluster; whereas variables vehicles involved, number of injuries, atmospheric factors, pavement markings and pavement width are identified only in one cluster.

Analysis of traffic accident severity using Decision Rules via Decision Trees

A Decision Tree (DT) is a potential method for studying traffic accident severity. One of its main advantages is that Decision Rules (DRs) can be extracted from its structure. And these DRs can be used to identify safety problems and establish certain measures of performance. However, when only one DT is used, rule extraction is limited to the structure of that DT and some important relationships between variables cannot be extracted. This paper presents a more effective method for extracting rules from DTs. The methods effectiveness when applied to a particular traffic accident dataset is shown. Specifically, our study focuses on traffic accident data from rural roads in Granada (Spain) from 2003 to 2009 (both included). The results show that we can obtain more than 70 relevant rules from our data using the new method, whereas with only one DT we would have extracted only five relevant rules from the same dataset.

Extracting decision rules from police accident reports through decision trees

Given the current number of road accidents, the aim of many road safety analysts is to identify the main factors that contribute to crash severity. To pinpoint those factors, this paper shows an application that applies some of the methods most commonly used to build decision trees (DTs), which have not been applied to the road safety field before. An analysis of accidents on rural highways in the province of Granada (Spain) between 2003 and 2009 (both inclusive) showed that the methods used to build DTs serve our purpose and may even be complementary. Applying these methods has enabled potentially useful decision rules to be extracted that could be used by road safety analysts. For instance, some of the rules may indicate that women, contrary to men, increase their risk of severity under bad lighting conditions. The rules could be used in road safety campaigns to mitigate specific problems. This would enable managers to implement priority actions based on a classification of accidents by types (depending on their severity). However, the primary importance of this proposal is that other databases not used here (i.e. other infrastructure, roads and countries) could be used to identify unconventional problems in a manner easy for road safety managers to understand, as decision rules.

Bayes classifiers for imbalanced traffic accidents datasets

Traffic accidents data sets are usually imbalanced, where the number of instances classified under the killed or severe injuries class (minority) is much lower than those classified under the slight injuries class (majority). This, however, supposes a challenging problem for classification algorithms and may cause obtaining a model that well cover the slight injuries instances whereas the killed or severe injuries instances are misclassified frequently. Based on traffic accidents data collected on urban and suburban roads in Jordan for three years (2009-2011); three different data balancing techniques were used: under-sampling which removes some instances of the majority class, oversampling which creates new instances of the minority class and a mix technique that combines both. In addition, different Bayes classifiers were compared for the different imbalanced and balanced data sets: Averaged One-Dependence Estimators, Weightily Average One-Dependence Estimators, and Bayesian networks in order to identify factors that affect the severity of an accident. The results indicated that using the balanced data sets, especially those created using oversampling techniques, with Bayesian networks improved classifying a traffic accident according to its severity and reduced the misclassification of killed and severe injuries instances. On the other hand, the following variables were found to contribute to the occurrence of a killed causality or a severe injury in a traffic accident: number of vehicles involved, accident pattern, number of directions, accident type, lighting, surface condition, and speed limit. This work, to the knowledge of the authors, is the first that aims at analyzing historical data records for traffic accidents occurring in Jordan and the first to apply balancing techniques to analyze injury severity of traffic accidents.

Development of safety performance functions for Spanish two-lane rural highways on flat terrain

Over decades safety performance functions (SPF) have been developed as a tool for traffic safety in order to estimate the number of crashes in a specific road section. Despite the steady progression of methodological innovations in the crash analysis field, many fundamental issues have not been completely addressed. For instance: Is it better to use parsimonious or fully specified models? How should the goodness-of-fit of the models be assessed? Is it better to use a general model for the entire sample or specific models based on sample stratifications? This paper investigates the above issues by means of several SPFs developed using negative binomial regression models for two-lane rural highways in Spain. The models were based on crash data gathered over a 5-year period, using a broad number of explanatory variables related to exposure, geometry, design consistency and roadside features. Results show that the principle of parsimony could be too restrictive and that it provided simplistic models. Most previous studies apply conventional measurements (i.e., R(2), BIC, AIC, etc.) to assess the goodness-of-fit of models. Seldom do studies apply cumulative residual (CURE) analysis as a tool for model evaluation. This paper shows that CURE plots are essential tools for calibrating SPF, while also providing information for possible sample stratification. Previous authors suggest that sample segmentation increases the model accuracy. The results presented here confirm that finding, and show that the number of significant variables in the final models increases with sample stratification. This paper point out that fully models based on sample segmentation and on CURE may provide more useful insights about traffic crashes than general parsimonious models when developing SPF.

Patterns of Single-Vehicle Crashes on Two-Lane Rural Highways in Granada Province, Spain: In-Depth Analysis Through Decision Rules

In Spain, 74% of injury crashes occur on rural two-lane highways. Therefore, one of the strategic priorities detailed in highway safety plans is the specific study of these highways. This study aimed at investigating crash patterns and contributory factors on rural two-lane highways so as to propose specific road safety countermeasures. The analysis method consisted in identifying decision rules extracted from decision trees (DTs). As the traditional method of rule extraction is limited by a DTs structure, some important relationships between variables may not be identified. For this problem to be overcome, an in-depth method for extracting rules from DTs was used. Because the implementation of any corrective road safety measure is constrained by the available resources, the strongest patterns that describe the road safety issue must be extracted. For identification of the strongest rules, a new criterion, lift increase criterion, was defined. Single-vehicle crashes on two-lane rural highways in the province of Granada in Spain were analyzed. Crash data were relative to the 7-year period 2003 to 2009. Rules were obtained by using both gain information and the information gain ratio as splitting criterion. The rules obtained by application of both criteria were consistent and complementary; therefore, the authors recommend the use of both methods to build DTs. Results of the study highlighted several patterns contributing to severe crashes and potentially effective countermeasures. Main patterns were pedestrian crashes, run-off-the-road crashes, run-off-the-road crashes involving powered two-wheelers, crashes involving powered two-wheelers, and crashes at night without illumination.

How to Expand Subway and Urban Railway Networks: Light Rail Extensions in Madrid, Spain

Residential areas of detached houses were built in north and east Madrid, Spain, during the last decade of the 20th century. Because the population density in those areas is low, implementing an efficient transport system is complicated. In south Madrid in the 1960s, however, huge commuter towns developed. Some of them were linked to a suburban network at some point, whereas others were far away. The Madrid region promoted the construction of several light rail lines to resolve this issue. The lines function as extensions to the existing rail network (subway or suburban trains). The determining factors that enabled these lines to be implemented and operated successfully are analyzed. The recommendations for implementing a light rail transit system, on the basis of that analysis, are: (a) the lines should not be very long; (b) they should have a segregated right-of-way; (c) the quality of service should be good; and (d) they should be coordinated with other modes of transport. Apart from public funding, they could be funded by additional resources collected from property tax and the concessionaire company (private funding). A concessionaire consortium may comprise construction companies, transport operators, financial institutions, rolling stock manufacturers, and consultancies. Finally, the concession should be granted for a 30- to 40-year period so that private stakeholders can recover their investment.

A proposal for cost-related and market-oriented train running charges

This paper examines some key aspects of a charging system for promoting railway transport, including charges reflecting a clear relationship with costs (transparency) and charges reflecting the quality of the infrastructure managers service. Train running charges recover track-related costs and can help to develop a charging system that meets these requirements. To orient train running charges to the market, a method for processing track maintenance and renewal costs is proposed whereby the quality of the service provided by an infrastructure is measured according to its utility to the railway undertaking. To achieve transparency, a single indicator is used for cost planning and the subsequent levying of costs on railway undertakings. The paper includes an example of how proposed train running charges would be calculated according to data from 14 European countries. The example shows that short-distance trains generate the lowest maintenance and renewal costs, followed by long-distance trains and freight trains.

Using Imprecise Probabilities to Extract Decision Rules via Decision Trees for Analysis of Traffic Accidents

The main aim of this study is focused on the extraction or obtaining of important decision rules (DRs) using decision trees (DTs) from traffic accidents’ data. These decision rules identify patterns related with the severity of the accident. In this work, we have incorporated a new split criterion to built decision trees in a method named Information Root Node Variation (IRNV) used for extracting these DRs. It will be shown that, with the adding of this criterion, the information obtained from the method is improved trough new and different decision rules, some of them use different variables than the ones obtained with the original method.