Developing Bicycle-Vehicle Crash-Specific Safety Performance Functions in Alabama Using Different Techniques

Analyzing angle crashes at unsignalized intersections using machine learning techniques

Using multivariate adaptive regression splines (MARS) to develop crash modification factors for urban freeway interchange influence areas

From the past to the future: Modeling the temporal instability of safety performance functions

The objective of this research was to develop new intersection crash prediction models for consideration in the second edition of the Highway Safety Manual (HSM), consistent with existing methods in HSM Part C and comprehensive in their ability to address a wide range of intersection configurations and traffic control types in rural and urban areas. The focus of the research was on developing safety performance functions (SPFs) for intersection configurations and traffic control types not currently addressed in HSM Part C. SPFs were developed for the following general intersection configurations and traffic control types: rural and urban all-way stop-controlled intersections; rural three-leg intersections with signal control; intersections on high-speed urban and suburban arterials (i.e., arterials with speed limits greater than or equal to 50 mph); urban five-leg intersections with signal control; three-leg intersections where the through movements make turning maneuvers at the intersections; crossroad ramp terminals at single-point diamond interchanges; and crossroad ramp terminals at tight diamond interchanges. Development of severity distribution functions (SDFs) for use in combination with SPFs to estimate crash severity as a function of geometric design elements and traffic control features was explored; but owing to challenges and inconsistencies in developing and interpreting the SDFs, it was recommended for the second edition of the HSM that crash severity for the new intersection configurations and traffic control types be addressed in a manner consistent with existing methods in Chapters 10, 11, and 12 of the first edition, without use of SDFs.

Accurately estimating annual average daily traffic (AADT) on minor roads is essential for assessing traffic-related air pollution (TRAP) exposure, particularly in areas where most people live. Our study assessed the direct and indirect external validity of three methods used to estimate AADT on minor roads in Melbourne, Australia.We estimated the minor road AADT using a fixed-value approach (assuming 600 vehicles/day) and linear and negative binomial (NB) models. The models were generated using road type, road importance index, AADT and distance of the nearest major road, population density, workplace density, and weighted road density. External measurements of traffic counts, as well as black carbon (BC) and ultrafine particles (UFP), were conducted at 201 sites for direct and indirect validation, respectively. Statistical tests included Akaike information criterion (AIC) to compare models' performance, the concordance correlation coefficient (CCC) for direct validation, and Spearman's correlation coefficient for indirect validation. Results show that 88.5% of the roads in Melbourne are minor, yet only 18.9% have AADT. The performance assessment of minor road models indicated comparable performance for both models (AIC of 1,023,686 vs. 1,058,502). In the direct validation with external traffic measurements, there was no difference between the three methods for overall minor roads. However, for minor roads within residential areas, CCC (95% confidence interval [CI]) values were −0.001 (−0.17; 0.18), 0.47 (0.32; 0.60), and 0.29 (0.18; 0.39) for the fixed-value approach, the linear model, and the NB model, respectively. In the indirect validation, we found differences only on UFP where the Spearman's correlation (95% CI) for both models and fixed-value approach were 0.50 (0.37; 0.62) and 0.34 (0.19; 0.48), respectively. In conclusion, our linear model outperformed the fixed-value approach when compared against traffic and TRAP measurements. The methodology followed in this study is relevant to locations with incomplete minor road AADT data.

Reliable estimates of heavy-truck volumes in the United States are important in a number of transportation applications including pavement design and management, traffic safety, and traffic operations. Additionally, because heavy vehicles emit pollutants at much higher rates than passenger vehicles, reliable volume estimates are critical to computing accurate inventories of on-road emissions. Accurate baseline inventories are also necessary to forecast future scenarios. The research presented in this paper evaluated three different methods commonly used by transportation agencies to estimate annual average daily traffic (AADT), which is used to determine vehicle miles traveled (VMT). Traffic data from continuous count stations provided by the Iowa Department of Transportation were used to estimate AADT for single-unit and multiunit trucks for rural freeways and rural primary highways using the three methods. The first method developed general expansion factors, which apply to all vehicles. AADT, representing all vehicles, was estimated for short-term counts and was multiplied by statewide average truck volumes for the corresponding roadway type to obtain AADT for each truck category. The second method also developed general expansion factors and AADT estimates. Truck AADT for the second method was calculated by multiplying the general AADT by truck volumes from the short-term counts. The third method developed expansion factors specific to each truck group. AADT estimates for each truck group were estimated from short-term counts using corresponding expansion factors. Accuracy of the three methods was determined by comparing actual AADT from count station data to estimates from the three methods. Accuracy of the three methods was compared using n-fold cross-validation. Mean squared error of prediction was used to estimate the difference between estimated and actual AADT. Prediction error was lowest for the method that developed separate expansion factors for trucks. Implications for emissions estimation using the different methods are also discussed.

Safety performance functions (SPFs) are essential tools to help agencies predict crashes and understand influential factors. Florida Department of Transportation (FDOT) has implemented a context classification system which classifies intersections into eight context categories rather than the three classifications used in the Highway Safety Manual (HSM). Using this system, regional SPFs could be developed for 32 intersection types (unsignalized and signalized 3-leg and 4-leg for each category) rather than the 10 HSM intersection types. In this paper, eight individual intersection group SPFs were developed for the C3R-Suburban Residential and C4-Urban General categories and compared with full SPFs for these categories. These comparisons illustrate the unique and regional insights that agencies can gain by developing these individual SPFs. Poisson, negative binomial, zero-inflated, and boosted regression tree models were developed for each studied group as appropriate, with the best model selected for each group based on model interpretability and five performance measures. Additionally, a linear regression model was built to predict minor roadway traffic volumes for intersections which were missing these volumes. The full C3R and C4 SPFs contained four and six significant variables, respectively, while the individual intersection group SPFs in these categories contained six and nine variables. Factors such as major median, intersection angle, and FDOT District 7 regional variable were absent from the full SPFs. By developing individual intersection group SPFs with regional factors, agencies can better understand the factors and regional differences which affect crashes in their jurisdictions and identify effective treatments.

Prediction of the shear modulus of municipal solid waste (MSW): An application of machine learning techniques

Calibration and development of safety performance functions for two-way stop-control intersections on rural two-lane highways in Louisiana

Land subsidence is a hazard that results from conditioning factors that cause environmental change and generate social and economic impacts. Some of these factors may increase dissolution of calcareous stones, change groundwater storage, or stem from mining, faulting, and seismic activity. Semnan Plain, Iran is experiencing land subsidence that, along with secondary impacts like surface fissures, is becoming increasingly troublesome. This study modeled land subsidence and created a susceptibility map using multivariate adaptive regression spline (MARS), mixture discriminant analysis (MDA), and boosted regression tree (BRT) machine-learning methods. Analysis of satellite imagery, documents reporting past subsidence, and a contemporary field survey identified 65 sinkholes on Semnan Plain. Twelve conditioning factors were selected for analysis from a review of the scholarly literature, field investigation, and data availability. The three methods were used to model subsidence from a training subset of the known sites. The models were validated with the remaining subset of subsidence locations. Finally, susceptibility maps were used to predict other sites that are highly likely to see subsidence. Receiver operating characteristic curves and the area under the curves (AUC) were applied to assess the accuracies of the maps. AUC values (0.637, 0.783, and 0.712 for the BRT, MARS, and MDA models respectively) showed that MARS generated the most accurate model, MDA generated the second most accurate, and BRT’s was the least accurate model. Subsidence susceptibility maps as produced here can be useful, meaningful, and functional tools for local and regional planners and policy makers involved in land use planning, resource management, and hazard mitigation.

ABSTRACTTransportation agencies rely on the Highway Safety Manual's (HSM) scientific predictive procedures to quantify road safety; however, the HSM methods may not be accurate for use by local jurisdictions. This study used data from rural two-lane, two-way intersections in South Dakota to compare jurisdiction-specific safety performance functions (SPFs) and HSM SPFs to determine which functions were more accurate when used locally. Jurisdiction-specific SPFs—without the use of a calibration factor—were found to be most accurate. The study also compared fixed severity proportion with a severity proportion function calibrated by local data. The authors assumed that the site-specific severity proportion would be more appropriate than a fixed value because the former reflects the relationship between severity proportion and site characteristics. However, results showed that the severity proportion function did not contribute significantly to prediction accuracy. The conclusion offers certain level of assurance for practitioners who are more likely to use a fixed injury severity proportion as a viable alternative to a site-specific severity proportion.

<p>The Highway Safety Manual uses safety performance functions (SPFs) and crash modification factors (CMFs) to quantify the safety effects of a roadway design and operational changes. A number of substantial methodological issues in SPF and CMF estimation exist that have not yet been completely addressed. Notably, although applicability of SPFs and CMFs may significantly vary by crash severity, only limited consideration is given to severity distribution. Moreover, when developing crash modification functions (CMFunctions), most of the current state-of-the-art studies only rely on the functional forms that may not address the non-linear effects of roadway attributes on safety. In addition, while many studies advocate using traffic conflicts as surrogate measures of safety to address some issues in CMF estimation, most of the current measures do not sufficiently take into account the severity of conflicts.</p> <p>The thesis is based on an amalgamation of four research papers that separately delve into these topics in Chapters 3 to 6. Chapter 3 develops and compares alternative approaches for predicting number of crashes for each severity level. The applicability of a two-stage SDF (severity distribution function) modeling approach is the main focus of this chapter. Chapter 4 applies Bayesian Markov Chain Monte Carlo (MCMC) simulation to investigate the functional forms used to derive CMFunctions in cross- sectional regression models. This chapter mainly focuses on the methodology and uses freeway median width as a case study. Chapter 5 aims to demonstrate the versatility of the approach by applying that to passing lanes on two-lane roads. The results of these two applications highlight the importance of using the functional forms that can capture non-linear effects of road attributes for CMF estimation in cross- sectional models, while providing robust CMFs for practical applications.</p> <p>Chapter 6 investigates a novel traffic technique using the data obtained from video observations at signalized intersections. The measures of post encroachment times and corresponding conflicting vehicle speeds are integrated to define a risk score which classifies conflicts into three severity levels. The results show strong relationships between the classified conflicts and both total and fatal/injury crashes and conform the promise of the approach for estimating CMFs where there are insufficient crash data for directly estimating them.</p>

Linear dependence of variables is a commonly used assumption in most diagnostic systems for which many robust methodologies have been developed over the years. In case the system nonlinearities are relevant, fault diagnosis methods, relying on the assumption of linearity, might potentially provide unsatisfactory results in terms of false alarms and missed detections. In recent years, many authors have proposed machine learning (ML) techniques to improve fault diagnosis performance to mitigate this problem. Although very powerful, these techniques require faulty data samples that are representative of any fault scenario. Additionally, ML techniques suffer from issues related to overfitting and unpredictable performance in regions which are not fully explored in the training phase. This paper proposes a non-linear additive model to characterize the non-linear redundancy relationships among the system signals. Using the multivariate adaptive regression splines (MARS) algorithm, these relationships are identified directly from the data. Next, the non-linear redundancy relationships are linearized to derive a local time-dependent fault signature matrix. The faulty sensor can then be isolated by measuring the angular distance between the column vectors of the fault signature matrix and the primary residual vector. A quantitative analysis of fault isolation and fault estimation performance is performed by exploiting real data from multiple flights of a semi-autonomous aircraft, thus allowing a detailed quantitative comparison with state-of-the-art machine-learning-based fault diagnosis algorithms.

With the rapid development of medicine and technology, machine learning (ML) techniques are extensively applied to medical informatics and the suboptimal health field to identify critical predictor variables and risk factors. Metabolic syndrome (MetS) and chronic kidney disease (CKD) are important risk factors for many comorbidities and complications. Existing studies that utilize different statistical or ML algorithms to perform CKD data analysis mostly analyze the early-stage subjects directly, but few studies have discussed the predictive models and important risk factors for the stage-III CKD high-risk health screening population. The middle stages 3a and 3b of CKD indicate moderate renal failure. This study aims to construct an effective hybrid important risk factor evaluation scheme for subjects with MetS and CKD stages III based on ML predictive models. The six well-known ML techniques, namely random forest (RF), logistic regression (LGR), multivariate adaptive regression splines (MARS), extreme gradient boosting (XGBoost), gradient boosting with categorical features support (CatBoost), and a light gradient boosting machine (LightGBM), were used in the proposed scheme. The data were sourced from the Taiwan health examination indicators and the questionnaire responses of 71,108 members between 2005 and 2017. In total, 375 stage 3a CKD and 50 CKD stage 3b CKD patients were enrolled, and 33 different variables were used to evaluate potential risk factors. Based on the results, the top five important variables, namely BUN, SBP, Right Intraocular Pressure (R-IOP), RBCs, and T-Cho/HDL-C (C/H), were identified as significant variables for evaluating the subjects with MetS and CKD stage 3a or 3b.

The cost of bridge construction is influenced by numerous internal and external factors, making it very difficult to approximate. Years prior to a bridge project letting, state highway agencies (SHAs) must set a reliable budget for the proposed project, but the information available to them is very limited. Developing reliable cost estimates for bridge projects during the early pre-construction phases is very important and challenging for SHAs. This study employed a non-parametric regression analysis technique—multivariate adaptive regression splines (MARS)—to model the conceptual cost of bridge projects. This novel approach does not require detailed construction documents and does not require strict assumptions to be valid for the developed model or to be reliable for the model predictions. MARS was applied to the empirical data gathered from a Florida Department of Transportation database. The 10-fold cross-validation method was employed in this study to assess model performance. The criteria to gauge overall model fit, generalizability, and prediction error were evaluated. The results revealed that the developed model consistently performed well, based on Cross-Validated R-square (CVRSq), Generalized Cross-Validation (GSV), Generalized R-square (GRSq), and max error.