A random forests approach to prioritize Highway Safety Manual (HSM) variables for data collection

Abstract

SUMMARYThe Highway Safety Manual (HSM) recommends using the empirical Bayes method with locally derived calibration factors to predict an agency's safety performance. The data needs for deriving these local calibration factors are significant, requiring very detailed roadway characteristics information. Many of these data variables are currently unavailable in most of the agencies' databases. Furthermore, it is not economically feasible to collect and maintain all the HSM data variables. This study aims to prioritize the HSM calibration variables based on their impact on crash predictions. Prioritization would help to identify influential variables for which data could be collected and maintained for continued updates, and thereby reduce intensive data collection efforts. Data were first collected for all the HSM variables from over 2400 miles of urban and suburban arterial road networks in Florida. Using 5 years (2008)(2009)(2010)(2011)(2012) of crash data, a random forests data mining approach was then applied to measure the importance of each variable in crash frequency predictions for five different urban and suburban arterial facilities including two-lane undivided, three-lane with a two-way left-turn lane, four-lane undivided, four-lane divided, and five-lane with a twoway left-turn lane. Two heuristic approaches were adopted to prioritize the variables: (i) simple ranking based on individual relative influence of variables; and (ii) clustering based on relative influence of variables within a specific range. Traffic volume was found as the most influential variable. Roadside object density, minor commercial driveway density, and minor residential driveway density variables were the other variables with significant influence on crash predictions.

PCA-based missing information imputation for real-time crash likelihood prediction under imbalanced data

“…Tree-based ensemble methods, such as the stochastic gradient boosting (SGB) and AdaBoost, also demonstrated strong ability in enhancing reliability of the real-time risk assessment (Ahmed and Abdel-Aty, 2013a). Motivated by the fact that a large number of explanatory variables might induce overfitting issues, random forest was widely used for selecting important factors (Saha et al, 2015). Kwak and Kho (2016) used the conditional logistic regression analysis to remove confounding factors, and developed separate models to predict crash occurrence with the genetic programming technique.…”

Tree-Based Classification of Tabla Strokes

“…It has been shown that random forest performs equally well or better than other methods on a diverse set of problems. It has been widely used in classification problems as diverse as bioinformatics 16 , medicine 17 , transportation safety 18 and customer behavior 19 .…”

Application of deep learning techniques in predicting motorcycle crash severity

“…For instance, Random Forest Model is a promising data mining approach employed in many studies to prioritize the variables associated with crashes. 6,13,14 Researchers also found the better prediction performance of the support vector machine (SVM) in developing crash injury severity models compared to other parametric models. [15][16][17] Moreover, despite limited application in transportation sector, multivariate adaptive regression splines technique has outstanding predictive power in crash injury analysis.…”

“…Machine learning algorithm has been used extensively in transportation safety studies. For instance, Random Forest Model is a promising data mining approach employed in many studies to prioritize the variables associated with crashes 6,13,14 . Researchers also found the better prediction performance of the support vector machine (SVM) in developing crash injury severity models compared to other parametric models 15–17 .…”