Safety Performance Functions have several important uses in road safety analysis. These functions, also known as Accident Prediction Models, are equations able to give an estimate of the expected average number of accidents at similar entities, relating the annual accident experience of an entity to its features. These safety tools can forecast the expected annual number of accidents for a given “past period” or “future period”, in way to allow the assessment of safety performance of an entity and the safety effect of design changes for new road projects and treatments to existing road. Moreover the application of these models can avoid the problems closely related with the police-reported accidents which are influenced by reportable criteria, police procedures, lack of reporting and errors to report. These models have to be properly calibrated, but this task is particularly hard due to the complexity to specify the mathematical form, the accommodation of the peculiarities of accident data and the transferability of models to other jurisdictions. The aim of this study is to develop a Safety Performance Function for four-legged signalized intersections; some of these are located on two collector roads, crossing build-up areas, while the others are in urban areas, so that all selected intersections are characterized by urban environment, factor that can directly influence the count of the expected accidents. The methodological approach used during this research lies on choosing an appropriate base model and on verifying its suitability to the real traits of the examined context. In order to obtain this purpose, it has been chosen one of the model for four-legged signalized intersections proposed during a research conducted in Toronto. It calculates the expected accidents for these types of entities as the product of the intersection traffic demands raised to a power; exactly the Toronto Safety Performance Function relates accidents to the entering AADT of the major and minor roads, whose exponents change with intersection features and type of accident data (injury or all accident severities). The selection of the more suitable model form has been based on the integral-derivative (ID) method. Basically the method consists of creating an empirical integral function (EIF), for each independent variable, and then to compare the EIF graph created with pre-established graphs of well-known functions (power, gamma, polynomial, etc…) in order to indicate the proper relationship between the dependent and independent variables. In order to obtain the coefficients of the selected statistical model it has been implemented a calibration procedure that, by using the method of the maximum likelihood function, assesses the model parameters that make this function the largest. The validity of the selected models and of its coefficients has been investigated with the Cumulative Residual (CURE) method: this method consists of plotting the cumulative residuals (the difference between the actual and fitted values for each intersection) for each independent variable. It is possible to assert that the selected Accident Prediction Model fits with good accuracy the data available as the cumulative residuals oscillate around the zero value and moreover lie between their two standard deviation boundaries (62σ*).

Adapting Safety Performance Functions for Signalized Four Legged Intersections

LOSA, MASSIMO;
2005

Abstract

Safety Performance Functions have several important uses in road safety analysis. These functions, also known as Accident Prediction Models, are equations able to give an estimate of the expected average number of accidents at similar entities, relating the annual accident experience of an entity to its features. These safety tools can forecast the expected annual number of accidents for a given “past period” or “future period”, in way to allow the assessment of safety performance of an entity and the safety effect of design changes for new road projects and treatments to existing road. Moreover the application of these models can avoid the problems closely related with the police-reported accidents which are influenced by reportable criteria, police procedures, lack of reporting and errors to report. These models have to be properly calibrated, but this task is particularly hard due to the complexity to specify the mathematical form, the accommodation of the peculiarities of accident data and the transferability of models to other jurisdictions. The aim of this study is to develop a Safety Performance Function for four-legged signalized intersections; some of these are located on two collector roads, crossing build-up areas, while the others are in urban areas, so that all selected intersections are characterized by urban environment, factor that can directly influence the count of the expected accidents. The methodological approach used during this research lies on choosing an appropriate base model and on verifying its suitability to the real traits of the examined context. In order to obtain this purpose, it has been chosen one of the model for four-legged signalized intersections proposed during a research conducted in Toronto. It calculates the expected accidents for these types of entities as the product of the intersection traffic demands raised to a power; exactly the Toronto Safety Performance Function relates accidents to the entering AADT of the major and minor roads, whose exponents change with intersection features and type of accident data (injury or all accident severities). The selection of the more suitable model form has been based on the integral-derivative (ID) method. Basically the method consists of creating an empirical integral function (EIF), for each independent variable, and then to compare the EIF graph created with pre-established graphs of well-known functions (power, gamma, polynomial, etc…) in order to indicate the proper relationship between the dependent and independent variables. In order to obtain the coefficients of the selected statistical model it has been implemented a calibration procedure that, by using the method of the maximum likelihood function, assesses the model parameters that make this function the largest. The validity of the selected models and of its coefficients has been investigated with the Cumulative Residual (CURE) method: this method consists of plotting the cumulative residuals (the difference between the actual and fitted values for each intersection) for each independent variable. It is possible to assert that the selected Accident Prediction Model fits with good accuracy the data available as the cumulative residuals oscillate around the zero value and moreover lie between their two standard deviation boundaries (62σ*).
9788890240997
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11568/95148
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact