See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329191384

Prediction of Driving Outcomes Based on Driver Behaviour and Roadway

Information

Article  in  Indian Journal of Pure and Applied Mathematics · February 2018

CITATIONS READS

0 152

2 authors:

Ybralem Bugusa Shruti Patil

Arba Minch University Symbiosis Institute of Technology
2 PUBLICATIONS   0 CITATIONS    11 PUBLICATIONS   6 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ybralem Bugusa on 26 November 2018.

The user has requested enhancement of the downloaded file.

International Journal of Pure and Applied Mathematics
Volume 118 No. 7 2018, 655-666
ISSN: 1311-8080 (printed version); ISSN: 1314-3395 (on-line version)
url: http://www.ijpam.eu
Special Issue
ijpam.eu

Prediction of Driving Outcomes Based on Driver

Behaviour and Roadway Information
YbralemBugusa Prof. Shruti Patil
Mtech Student, Department of Computer Science and Assistant Professor, Department of Computer Science and
Engineering Engineering
Symbiosis International University, Pune, India Symbiosis International University, Pune, India
ybralem.wekele@sitpune.edu.in shruti.patil@sitpune.edu.in

Abstract- Intelligent Transportation System (ITS) is advanced
transportation framework for the more quick-wittedutilization of
transportation system. The setup consists of an arrangement of
coordinates propelled advancements which utilize hardware,
computers, correspondence, sensor, and cameras to upraise
travelers with relevant information for the safety and competence
of transportation system. Advanced Driver Assistance System
(ADAS) is a piece of ITS, that helps drivers in the driving
procedure to empower safe condition. In such a system,to
guarantee the wellbeing of all the street clients, it is important to
upgrade the execution of Advanced Driver Assistance System
(ADAS) for lower classes of vehicles. Real-time driving safety risk
prediction is a primary component of an ADAS.In this paper, we
are going to use a dataset from Secondary Strategic Highway
Research Program (SHRP2) Naturalistic Driving Study. The
elastic net regularized multinomial logistic regression model is
going to be utilized to predict the traffic safety risk that can be
updated to the individual driver based on specific variables. The
prediction performance of the model will be improved by using
re-sampling techniques.
Keywords: traffic safety, driving outcome, naturalistic driving
behavior, multinomial logistic regression, elastic net, re-sampling,
advanced driver assistance system.
Developing a model that predicts the real-timeroad safety risk
(driving outcome) for ADAS are important in creating a safer
driving environment for drivers, passengers, and pedestrians.
For all of this, it is important to have a real-time data of the
traffic safety factors,i.e., vehicle and other infrastructure
device information with regard time, location and other
relevant information in association with the main safety
factors of driver behavior, roadway information, and
invironmental condition data. The aim of this paper is
developing model that predicts the real-time driving outcome
of driver during driving situation that can be updated to
individual driver information. It will use Strategic Highway
Research Program 2(SHRP2) Naturalistic Driving Study
(NDS) data to make a predictive model of real-time traffic risk
for ADAS. The model uses some of the data above by
aggregating the data of individual trip.
To classify the driving outcome of the driver using multiple
traffic safety factor, the elastic net regularized multinomial
logistic regression [1] will be used due to its ability work with
correlated variable and bias/variance trade-off.

I. INTRODUCTION The remaining of this workis arranged as follows: First, we

provide the related work with a brief description and evidence
Having the growing technologies in intelligent transportation,
about SHRP 2 data and the influence of driver behavior in
improving the performance of Advanced Driver Assistance
traffic safety risk prevention. Finally, we are going to discuss
System (ADAS) is necessary to ensure road safety. Modern
proposed work and conclusion.
ADAS are systems that help drivers in real time driving
process by warning to the driver or actuation of the control
systems to reduce traffic collisions that cause injury, death of
pedestrians as well as drivers and property damage.

655
International Journal of Pure and Applied Mathematics
II. LITERATURE REVIEW
a. Naturalistic Driving Study data
Traffic damages are an expanding group medical problem, too
much touching touchy groups of street and poor people.
Moreover, half of the general population executed in vehicle
accidents are youthful grown-ups elderly in the age of 15 and
44 years – regularly the representatives in a family and the
cutting edge the next generation which will receive their
country. Furthermore, Traffic accident Cost developing
countries 1% to 2% of their gross national commodity– more
than the aggregate development of these countries. [21]
To overcome this distraction many analysis and crash risk
level prediction model have been done by evaluating the crash
casual factors data to provide vital information for effective
transportation policy, vehicle design, driver awareness and
potential countermeasures. However, there have been certain
criticized in data observation studies of crash casual factors
due to poor quality data; absence of available and valuable
exposure measures linked to the observation; the
incomparability of self-revealed safety-related events; and, the
difficulty in evaluating accountability for the safety-related
events [2].
[5] Paper uses traffic data from loop detector to develop
prediction models of real-time crash risk for different segment
types and states of traffic flow on Korean expressways. They
address the result difference in occupancy and level of
overcrowding play a significant role in the ramp vicinities and
level of congestion difference between the downstream and
upstream leads to a hazardous condition. However, it has the
656
Special Issue
limitation of lack of weather-related data and lane-based
traffic data.
[12] paper takes into account the traffic data from loop
detector data and weather data from web crawler with the
objective of investigating predictors of future traffic crash
involvement.They handle bias of data for former drivers and
compare characteristics of old drivers and crash-involved
drivers. In the study, it includes only older adults living in
Kashiwa. Therefore, the sample is biased toward active,
healthy older adults who could take part in social
activities.The sample size is small.
The second Strategic Highway Research Program (SHRP2) is
research method studying the largest and most comprehensive
Naturalistic Driving Study (NDS) ever undertaken linked with
Roadway Information Database (RID). Initially, the 100-Car
Naturalistic Driving Study collect the large-scale NDS data in
the US which is first instrumented-vehicle study funded by the
National Highway Traffic Safety Administration (NHTSA)
and the Virginia Department of Transportation (VDOT). Then
it is followed by a larger and more wide-ranging study, the
Strategic Highway Research Program 2 (SHRP2) conducted
from 2006 to 2015 holds data about driver behavior, road,
vehicle, and weather and traffic conditions with age 16-68 in
the event of either crash or near-crash.
The NDS and RID data allow researchers to associate driving
behavior to specific roadway conditions for better analysis and
traffic risk prediction. Detail report of SHRP2 NDS data
instruments, data dictionary, all variables of the traffic safety
factors and how to query and filter states in [11] [18-19]
papers.
International Journal of Pure and Applied Mathematics
Fig 1: SHRP2 NDS dataset structure
In this study, we are going to use the SHRP2 NDS particular
variable of the safety factors,i.e., driver behavior, road,
vehicle, and whether information due to several reasons [2]:
 Have advanced data collection techniques for
different safety factor
 No information biases and
 Quality information primarily to identify the driver
fault in near-crash/crash events.
b. Driver behavior
Recently, many kinds of researchwork in the prediction of
real-time risk to prevent the sudden loss of human life by car
and other vehicle and acknowledged that driver behavior plays
a significant role in driving risk [1]. Itpresents that nearly 90%
of the light-vehicle crashes included a similar sort of human
error such as impaired conditions, accidental mistakes, and
risky driving behavior and driver mistake is additionally a
center explanation behind roughly 87% of every single
business vehicle crashes. Since factors associated with
individual driver is different [10], developing models that can
be customizable to the individual driver is essential to making
good countermeasures to traffic safety risk. Driving behavior
can be more influential when it is associated with roadway,
vehicle and weather condition data.
In [5] paper the authors develop real-time crash risk prediction
models for different segment types and traffic flow states on
Korean expressways using Genetic programming technique.
657
Special Issue
Authors [20] investigate the potential contribution of three
predictors (i.e., driver anger, impulsiveness, and
aggressiveness) of aggressive and transgressive manners on
the road using multiple regression analysis models.
The [6] develop a model for detecting the driver‟s
involvementlikelihood in secondary tasks from distinctive
attributes of driving behavior (i.e., lateral acceleration,
longitudinal acceleration,throttle position, and yaw rate).
Using supervised feed-forward Artificial Neural Network
(ANN) finds that the Selected driving performance attributes
were useful in detecting the associated secondary tasks with
driving behavior
In [14] the author uses driver behavior and physiological
features for crash prediction to advanced vehicle collision
avoidance system using supervised learning model.
Several studies use different data set to explore the
relationship between the involvement and the traffic safety-
related events and considers input data used for their work.
However, due to several reasons they aware some
improvements on data sample usage, some of them they don‟t
useweather-related data and lane-based traffic data [5]. Others
they may not evaluate variables that influence the performance
of prediction [20] (i.e., years of driving experience is not
considered).
In this study, we are going to use the driver behavior as a
major contributor to safety risk with external factors that may
influence it,i.e., roadway information weather condition, etc.
International Journal of Pure and Applied Mathematics Special Issue

of SHRP2 NDS data which is the vast research method ever
conducted. It contains data from vehicles, road, and weather
condition data about driver behavior. We will consider some
specific variables from all this data that are important for the
predictor to classify the real-time driving the outcome of the
individual driver.
Many studies have been conducted by using driver behavior
with different factors to predict
safety-related traffic problems. There are also studies which
donot consider other factors rather than driver behavior.

TABLE 1: FACTORS USED TO PREDICT TRAFFIC SAFETY RISK

Paper Factors used to predict trafficsafety risk Classification Algorithm Accuracy

[1] Driver behavior along with roadside information Elastic net 68.5%

[2] Vehicle ROR safety-related data Bayesian MultivariatePoissonLog 97%

Normal Model
[3] Driver behavior, vehicular and road data are used Classification and Regression Tree 66%
(CART)
[6] Vehicle-speed, vehicle longitudinal and lateral Artificial Neural Network (ANN) 99.5%
acceleration, yaw rate, and throttle position of the
vehicle to predict secondary task of the driver.
[9] Combination of driver glance behavior and traffic Linear Regression -
density
[10] It uses personality, demographic and driving behavior Negative Binomial (NB) Model and -
data Logistic Regression
[12] Driver behavior and crash history A Multinomial Logistic Regression -

[14] Driver behaviors and vehicle information Discriminant Analysis 95.25%

Paper [3], [6], [9] and [10]‟s has considered the additional Paper [1] uses weather and vehicle information, in addition to
factors like vehicle information and other factors. But, they driver behavior and roadside information which can be more
have used a limited number of sample data which can affect accurate in identifying safety-related traffic problems.
the accuracy of the system. In paper [14], the driver data was
collected from simulator driving. In this case, the accuracy can
be less when it is compared to thereal driving situation.

Fig 2: Factors affecting traffic safety risk

658
International Journal of Pure and Applied Mathematics Special Issue

To improve the accuracy of [1], we will usere-sampling consists of 19,600 baseline events with 27 variables which
technique. includes detailed epochs, driver state, and driving environment
information derived from video reduction. Event video
III. PROPOSED APPROACH
reduced data consists of 68 crashes or 760 near-crashes with
57 variables which includes detailed event, driver state, and
a. Data preparation and sampling
driving environment information derived from video
reduction. By joining the dataset into Baseline, Near-crash and
For further pre-processing of the SHRP2NDS data set, we will
Crash events, we have selected total 21 important variables
use Data munging and wrangle for cleaning and punctuation
which includes driver state, and driving environment
correction. It also helps to check the structure and value of the
information. Hence the number of Baseline event data is
data according to the data dictionary; thiscontributes to map
extremely large. Therefore, we have applied random sampling
the data into R. R is the tool that we will use in this study. To
techniques on major class (i.e. baseline event data) to reduce
handle the missing value, we are going to use Random Forest
the data size required to achieve better classification.After data
package to fill the missing value which is the fastest approach
cleaning, 3353 total data are left which consists of 2595-
[26].Multivariate Imputation by Chained Equations (MICE) is
Baseline (B), 969-NearCrash (N) and 62- Crash (C)Event
an expensive method used to handle the missing value [27].
Datasets.
We have collected the data from baseline video reduced data
and Event video reduced data.Baseline video reduced dataset
3000
2595
2500
Number of Data

2000

1500

1000 696
500
62
0
B N C

Events

Fig 3: Reduced dataset

Various applications can overcome the problem of imbalanced
data between classes. The rightness of prediction can be
improved when it handle the imbalanced data. The occurrence
of imbalanced data is generalized to one-sided towards the
larger part class. In any case, in a few applications, the
accuracy of prediction in minority class is additionally
noteworthy as much as in majority class. The accuracy in
minority class will be increased by using re-sampling
techniques. Such as under-sampling, over-sampling and
bootstrapping are the most used techniques. Under-sampling
and over-sampling are re-sampling based methods which don't
consider how the data are scattered in the space [25].
Bootstrapping is statistical resampling strategy to appraise the
accuracy of the sample by utilizing subsets of accessible data
or drawing haphazardly with substitution from a set of data
points or original data.
In this study, we are going to use bootstrapping method to
increase the accuracy of the classifier and overcome such
problem.
b. Driving outcome classification
The driving outcome or risk events of SHRP2 NDS are into
three crash, near-crashand baseline or normal state. Near-crash
is conflict situation requiring a rapid and careful maneuver to
avoid an accident.
The author [1] further decomposes the events into 5 class or
outcome to better create a safe environment for pedestrians as

659
International Journal of Pure and Applied Mathematics Special Issue

well as driver based on the unsafe driver behavior and decomposed the dataset into Baseline1, Baseline2, Baseline3,
secondary task that the driver involved. Accordingly, we have Near-crash and Crash.
TABLE 2: THE DECOMPOSED OUTCOME AND THEIR MEANING

No. Outcome If it is

1 Baseline1(normal driving1) Minimum risk

2 Baseline2(normal driving2) Self-hazardous

3 Baseline3(normal driving3) Hazardous to self and others

4 Near crash Near crash

5 Crash Crash

The risk events were three which are showing by color

spectrum for normal (baseline) state is “green,”for near crash
“yellow” and “red” for crash event.

3000
2595
2500
Number of Data

2000

1500

1000
696

500
62
0
B N C

Events

Fig4: Driving outcome before classification

To create safer environment which is especially usefulfor driving slowly in relation to other vehicle) and secondary
young drivers as well as drivers with limited driving ability. task(i.e. texting, cellphone call, eye glance) that the driver
The baseline event is divided into three based on the driver involves during driving situation.
involvement in unsafe driving behavior (i.e. improper turn,

660
International Journal of Pure and Applied Mathematics Special Issue

Fig5: Driving outcome architecture

This can be done by selecting the prevailing unsafe driving The following figure illustrates the number of each driving
behavior and secondary task from the dataset. The outcomes after classification.
decomposed one is show in table2, two colors are added to
represent the level of risk baseline1 is “green” as it
is,baseline2 is “green-yellow,” and baseline3 is “khaki.” The
remaining near-crash and crash event color is as it is. For
more,we have shown in below figure 5.

Driving Outcome
1600
1344
1400
1200
1000
800 711 696
540
600
400
200 62
0
Baseline1 Baseline2 Baseline3 Near-Crash Crash

Fig 6: Driving outcome after classification

661
International Journal of Pure and Applied Mathematics
In this paper, we propose weighted regularized multinomial
logistic regression (MLR) which is precisely known elastic net
[23] for variable selection and classification.
Multinomial Logistic Regression is the type of logistic
regression technique used to evaluate the parameters of a
multivariate explanatory model used when the dependent
variable is binary (0/1, true/false and yes/no) and the
independent variable is continuous or categorical [8].
It is the type of logistic regression with more than two
outcomes but not ordinal. Hence we have more than two
outcomes we will use the multinomial logistic regression.
However, it's hard to work with a highly correlated
independent variable to overcome it will collaborate with a
regularized method that is weighted regularized multinomial
logistic regression. Elastic net is a merging of L1 and L2
regularization.
LASSO regression is the type of regression technique that
works with highly correlated independent variable solves the
high variable relation by shrinkage parameter lambda which
takes one variable and shrinks the rest to zero that is the
problem with multicollinearity. It is the L1 regularization [23-
24].
Fig 7: Architecture of proposed work.
662
Special Issue
In reverse, the Ridge regression is L2 regularization solves the
multicollinearity by shrinking the value of coefficients but
doesn‟t reach to zero [23-24].
Elastic net is a group of Lasso and Ridge Logistic Regression.
While we are working with high dimensional, correlated
features, other reasons that improve the accuracy and
performance of prediction are:
Reasons why we will use elastic net:
 Integrated variable selection,
 Capability to work with bias/variance trade-off,
 Cost-sensitive classification,
 Capacity to handle highly correlated variables,
 Ability to deal with high-dimensional problems, and
 Work with both numerical and categorical variables
Problems in prediction can happen due to either biased or
variance; It may also occur by both. Regularized regression
techniques have their own mechanism to overcome this:
International Journal of Pure and Applied Mathematics Special Issue

TABLE 3: FORMULATION OF ELASTIC NET

Ridge Regularized
Advantage  Allow where n < p
 Solve multi collinearity by minimizing coefficients but not to zero.
 L2 norm based penalty.
Disadvantage  No feature selection.
Formula 𝑛 𝑘 𝑘

𝛽 𝑟𝑖𝑑𝑔𝑒 = min (𝑦𝑖 − 𝛽𝑜 − 𝑢𝑖𝑗 𝛽𝑗 )2 + 𝜆 𝛽𝑗2

𝛽
𝑖=1 𝑗 =1 𝑗 =1

𝑛 𝑘 𝑘

𝛽 𝑟𝑖𝑑𝑔𝑒 = min (𝑦𝑖 − 𝛽𝑜 − 𝑢𝑖𝑗 𝛽𝑗 )2 𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝛽𝑗2 ≤ 𝑡

𝛽
𝑖=1 𝑗 =1 𝑗 =1

Lasso Regularized
Advantages  Allow where n < p
 Similar with ridge but multi collinearity by shrinking coefficients to zero.
 L1 norm based penalty.
 Feature selection

Disadvantages  Cannot perform group selection

Formula 2
𝑛 𝑘 𝑘
𝑙𝑎𝑠𝑠𝑜
𝛽 = min 𝑦𝑖 − 𝛽𝑜 − 𝑢𝑖𝑗 𝛽𝑗 +𝜆 𝛽𝑗
𝛽
𝑖=1 𝑗 =1 𝑗 =1
𝑛 𝑘 𝑘

= min (𝑦𝑖 − 𝛽𝑜 − 𝑢𝑖𝑗 𝛽𝑗 )2 𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝛽𝑗 ≤ 𝑡

𝛽
𝑖=1 𝑗 =1 𝑗 =1

Elastic Net Regularized

Advantages  Allow where n < p
 It is shrinkage method.
 Uses both L1 and L2 norm based penalty.
Formula 2
𝑛 𝑘 𝑘 𝑘

= min 𝑦𝑖 − 𝛽𝑜 − 𝑢𝑖𝑗 𝛽𝑗 + 𝜆2 𝛽𝑗2 + 𝜆1 𝛽𝑗

𝛽
𝑖=1 𝑗 =1 𝑗 =1 𝑗 =1

2
𝑛 𝑘

= min 𝑦𝑖 − 𝛽𝑜 − 𝑢𝑖𝑗 𝛽𝑗 + 𝜆2 𝜷 2 + 𝜆1 𝜷 1
𝛽
𝑖=1 𝑗 =1

Note: 𝜆1 = 0, 𝜆2 > 0 is equivalent to ridge regression and

𝜆1 > 0, 𝜆2 = 0 is equivalent to the LASSO. If both 𝜆1 >
0, 𝜆2 > 0it is Elastic Net.
TABLE 4: COMPARISON OF ELASTIC NET CLASSIFIER WITH OTHER.

Ridge Logistic Lasso Logistic

Features/classification algorithm SVM Elastic net
Regression Regression

Performing automatic Variable selection NO NO Can„t complete group Yes

selection

663
International Journal of Pure and Applied Mathematics Special Issue

Handling of Bias/variance trade-off NO Yes Yes Yes

Dealing with Correlated data and high NO NO NO YES

dimension data

Elastic net is supervised classification model used in many  5 = crash

𝑒𝑙𝑎𝑠𝑡𝑖𝑐
areas for variable selection, optimization and classification o 𝛽 = argmin𝛽 (||𝑦 − 𝑋𝛽||22 +
purpose [22] [24]. 2
𝜆2 𝜷 + 𝜆1 𝜷 1 )
o Naïve elastic net penalty
𝑝 1
 𝑃𝛼 = 𝑖=1 [2 1 − 𝛼 𝑏𝑗2 + 𝛼|𝑏𝑗 |]

𝑓𝑜𝑟 𝑒𝑙𝑎𝑠𝑡𝑖𝑐 𝑛𝑒𝑡, 0 < 𝛼 <
Elastic net prediction model for driving outcome 1
 But, it has the problem of double
Suppose: driver i, trip j, location l and time t. shrinkage, with the same α level
o To overcome this problem, elastic net uses,
Procedure:
 Penalty = 1 − 𝛼 𝛽 2 + 𝛼|𝛽|2
𝜆2
1. Yijkl={1,2,3,4,5};  = 𝜆2 |𝛽|2 + 𝜆1 |𝜷|1 where 𝛼 =
𝜆 1 +𝜆 2
o 1,2,3,4 and 5 indicates the driving outcome.
 𝛽 𝑒𝑙𝑎𝑠𝑡𝑖𝑐 =
Such as Baseline1, Baseline2, Baseline3, Near-
crash and Crash respectively. (1 + 𝜆2 )𝛽 (𝑛𝑎𝑖𝑣𝑒 𝑒𝑙𝑎𝑠𝑡𝑖𝑐 𝑛𝑒𝑡)
2. Consider the independent safety factors represented by
Zijkl
o Zijkl, includes the target driver, vehicle, and all IV. CONCLUSION
other factors such as roadway information,
historical crash events, weather information. Prediction of road traffic risk during driving process plays an
3. Hence, the function f(ZijklYijkl), shows the relationship important role in saving of human life as well as infrastructure
between independent (predictor) and dependent damage. Since, driving outcome of the driver can be
(driving outcome) variables influenced by internal and external factors, it is being better to
4. Divide the safety variables based on the regularity of develop predictive model that can be customizable for
real-time changes: individual driver. In this investigation, we haveoffered elastic-
o Time-series variables, Tij(l,L),(k,N) , net regularized multinomial logistic regressionfor prediction
o Event variables, evij(l,L),(k,N) . Where
and classification of driving outcome and we are going to use
 T = constant time step,
the SHRP2 NDS dataset to achieve this. As improvements, we
 N = length of time-series
are going to use re-sampling technique to improve the
5. Redefine location and time of the predictor as
Zij(l,L),(k,N) performance of the model. For further extension of this work,
6. Define the conditional probability of the driving using best search algorithm and advanced machine learning
outcome as the following: algorithm are possible to improve its performance.
o Poutcome = P(Yijkl=outcome | Zij(l,L),(k,N)) =
f(Zi,jij(l,L),(k,N)) REFERENCES
[1] Arbabzadeh, Nasim, and Mohsen Jafari. "A Data-Driven Approach for
7. The driving outcome to be predicted given Zij(l,L),(k,N)
Driving Safety Risk Prediction Using Driver Behavior and Roadway
computed as: Information Data." IEEE Transactions on Intelligent Transportation
o Yijlt= Arg max Poutcome Systems (2017).
Outcome=1, 2, 3…number of class
[2] Wu, Kun-Feng, Jonathan Aguero-Valverde, and Paul P. Jovanis. "Using
8. Based on above formulation, we can apply elastic-net naturalistic driving data to explore the association between traffic safety-
model to this work. related events and crash risk at the driver level." Accident Analysis &
Prevention 72 (2014): 210-218.
o Number of class k = 5,
[3] Wang, Jianqiang, et al. "Driving risk assessment using near-crash
o Outcome database through data mining of tree-based model." Accident Analysis &
 1 = baseline 1, Prevention 84 (2015): 54-64.
 2 = baseline 2, [4] Li, G., Li, S. E., Cheng, B., & Green, P."Estimation of driving style in
 3 = baseline 3, naturalistic highway traffic using maneuver transition probabilities."
 4 = near-crash, Transportation Research Part C: Emerging Technologies 74 (2017):
113-125.

664
International Journal of Pure and Applied Mathematics Special Issue

[5] Kwak, Ho-Chan, and Seungyoung Kho. "Predicting crash risk and [23] Friedman, Jerome, Trevor Hastie, and Rob Tibshirani. "Regularization
identifying crash precursors on Korean expressways using loop detector paths for generalized linear models via coordinate descent." Journal of
data." Accident Analysis & Prevention 88 (2016): 9-19. statistical software 33.1 (2010): 1.
[6] Ye, M., Osman, O. A., Ishak, S., & Hashemi, B."Detection of driver [24] Boulares, Mehrez, and Mohamed Jemni. "Learning sign language
engagement in secondary tasks from observed naturalistic driving machine translation based on elastic net regularization and latent
behavior." Accident Analysis & Prevention 106 (2017): 385-391. semantic analysis." Artificial Intelligence Review 46.2 (2016): 145-166.
[7] You, Jinming, Junhua Wang, and Jingqiu Guo. "Real-time crash [25] Thanathamathee, Putthiporn, and ChidchanokLursinsap. "Handling
prediction on freeways using data mining and emerging techniques." imbalanced data sets with synthetic boundary data generation using
Journal of Modern Transportation (2017): 1-8. bootstrap re-sampling and AdaBoost techniques." Pattern Recognition
[8] Starkweather, Jon, and Amanda Kay Moske. "Multinomial logistic Letters 34.12 (2013): 1339-1347.
regression."Consulted page at September 10th: http://www. unt. [26] Liaw, Andy, and Matthew Wiener. "Classification and regression by
edu/rss/class/Jon/Benchmarks/MLR_JDS_Aug2011. pdf 29 (2011): randomForest." R news 2.3 (2002): 18-22.
2825-2830. [27] Buuren, Stef, and Karin Groothuis-Oudshoorn. "mice: Multivariate
[9] Tian, R., Li, L., Chen, M., Chen, Y., & Witt, G. J. "Studying the imputation by chained equations in R." Journal of statistical software
effects of driver distraction and traffic density on the probability of 45.3 (2011).
crash and near-crash events in naturalistic driving environment." IEEE
Transactions on Intelligent Transportation Systems 14.3 (2013): 1547-
1555.
[10] Guo, Feng, and Youjia Fang. "Individual driver risk assessment using
naturalistic driving data." Accident Analysis& Prevention 61 (2013): 3-
9.
[11] Brewer, Marcus A., Shannon Barkwell, Pei-Sung Lin, SeckinOzkul,
Walter R. Boot, Priyanka Alluri, Larry T. Hagen et al. "Exploration of
the SHRP2 naturalistic driving study data to identify factors related to
the selection of freeway ramp design speed." Ann Arbor 1001 (2017):
48109-2150.
[12] Kosuge, R., Okamura, K., Kihira, M., Nakano, Y., & Fujita,
G."Predictors of driving outcomes including both crash involvement
and driving cessation in a prospective study of Japanese older drivers."
Accident Analysis & Prevention 106 (2017): 131-140.
[13] Klauer, S. G., Dingus, T. A., Neale, V. L., Sudweeks, J. D., & Ramsey,
D. J. "The impact of driver inattention on near-crash/crash risk: An
analysis using the 100-car naturalistic driving study data." (2006).
[14] Ba, Y., Zhang, W., Wang, Q., Zhou, R., & Ren, C."Crash prediction
with behavioral and physiological features for advanced vehicle
collision avoidance system." Transportation Research Part C:
Emerging Technologies 74 (2017): 22-33.
[15] Murray, Daniel C., Brenda Lantz, and Stephen A. Keppler. "Predicting
truck crash involvement: Developing a commercial driver behavior-
based model and recommended countermeasures." (2005)
[16] Treat, John R., et al. "Tri-level study of the causes of traffic accidents:
final report. Executive summary." (1979).
[17] Fell, J. C., D. L. Hendricks, and M. Freedman. "The relative frequency
of unsafe driving acts in serious traffic crashes." PROCEEDINGS OF
THE 44TH ANNUAL CONFERENCE OF THE ASSOCIATION FOR
THE ADVANCEMENT OF AUTOMOTIVE MEDICINE, CHICAGO,
ILLINOIS, USA, OCTOBER 2-4, 2000. 2000.
[18] Dingus, Thomas A., Feng Guo, Suzie Lee, Jonathan F. Antin, Miguel
Perez, Mindy Buchanan-King, and Jonathan Hankey. "Driver crash risk
factors and prevalence evaluation using naturalistic driving data."
Proceedings of the National Academy of Sciences 113, no. 10 (2016):
2636-2641.
[19] Files, Trip. "Analyzing Driver Behavior Using Data from the SHRP 2
Naturalistic Driving Study." (2013).
[20] Berdoulat, Emilie, David Vavassori, and María Teresa Muñoz Sastre.
"Driving anger, emotional and instrumental aggressiveness, and
impulsiveness in the prediction of aggressive and transgressive
driving." Accident Analysis & Prevention 50 (2013): 758-767.
[21] Peden, Margie. "World report on road traffic injury prevention."
(2004).
[22] Li, Q., Xie, B., You, J., Bian, W., & Tao, D."Correlated logistic model
with elastic net regularization for multilabel image classification."
IEEE Transactions on Image Processing 25.8 (2016): 3801-3813.

665
 666

View publication stats