Sei sulla pagina 1di 5

Global Journal of Advanced Engineering Technologies

ISSN (Online): 2277-6370 & ISSN (Print):2394-0921

Volume 4, Issue 3- 2015

ROAD FATALITIES IN KUWAIT:


TRENDS, PREDICTION AND STRATEGIES
1

Jamal AlMatawah1 , Khair Jadaan2


Assistant Professor, Civil Engineering Department, College of Technological Studies, Kuwait (PAAET)
2
Professor, Civil Engineering Department, University of Jordan

Abstract: Road traffic accidents and their resulting


fatalities are regarded as a growing social and
economic problem especially in developing countries
where population and vehicle ownership are increasing
rapidly while many featured necessary to reduce
accident and fatality rates are not being introduced
effectively.
Estimates of the expected number of accidents are
necessary for the identification of sites which require
safety treatment and evaluation of such treatments; yet,
limited work, only of aggregate nature, was carried out
on accident predictions in Kuwait. A more accurate
estimate of accidents would enable a better selection of
appropriate potential countermeasures.
This paper describes the development of a prediction
model for estimating road fatalities in Kuwait
incorporating variables related to road, traffic and
environment. The data required for model development
was obtained from the Kuwaiti General Traffic
Department, Public Authority for Civil Information,
and Ministry of planning (Annual Statistical Abstract)
for the years 1997 through 2014. The variables that
were found significant are population and number of
registered vehicles with the models coefficient of
determination (R2) being 0.803, but the length of paved
roads was not significant. The paper also reviews
various potential countermeasures that are believed to
combat the problem, and describes the details of a
National Road Safety Plan recommended to improve
traffic safety in Kuwait.
Keywords:
Accident
Trends,
Accident/fatality
Prediction Models, Countermeasures, Road Safety
Strategies
I. INTRODUCTION
Road accidents and their consequences constitute a
growing social and economic problem in developing
countries. However, reliable data on vehicle and people
travel are not normally available and the commonly
used measures of risk exposure are therefore the
number of motor vehicles of different types, and
population [7].
Accident prediction models (APMs) are used for
estimating the expected number of accidents on urban
transportation network such as arterial roads,
roundabouts and intersections. The predicted number
of accidents is then used as a more stable estimate of
the number of accidents, instead of the observed
number at a single facility.
Various researches have been carried out to investigate
and predict accidents in developing countries using
www.gjaet.com

time-series data of road fatalities [6],[7]. Macro-level


models were also developed [5].
Arndt and Trout beck [2] applied multiple linear
regression analysis to determine the variation in the
number of accidents at roundabouts as a function of the
roundabout geometry variables and accident rates in
rural and urban areas. Baek et al. [3] presented a
method to estimate the relationship between accidents
and trip length on freeways based on the concept that
OD (Origin and Destination) trips include travel
behaviour, namely travel distance, in the freeway
network. Nambuusi et. al, and Marchionna et. al,
[12],[9] developed accident prediction models for
signalized intersections based on intersection
characteristics, environmental factors, and accident
history. More recently, Al-Matawah and Jadaan [1]
investigated the application of prediction techniques to
road safety in developing countries and produced a
prediction model for traffic fatalities using a regression
analysis of time-series fatality data. Similar work was
carried out using data from the UAE, Jordan and Qatar
[4].
Mohamed [11] developed econometric models for road
accidents through time series and cross sectional
modelling. The empirical part of this research involved
estimating accident risk function using ARIMA models,
regression models with ARMA errors, and logistic
regression for the explanation and prediction of
fatalities. Artificial Neural Network (ANN) was also
used for the same purpose of predicting fatalities [8]
This paper describes the development of a prediction
model for estimating road fatalities in Kuwait as a
function of demographic, traffic, and road network
variables. Some potential countermeasures together
with a recommended road safety strategy to improve
road safety in Kuwait are outlined.
II. TREND ANALYSIS OF THE NUMBER AND
RATES OF ACCIDENTS
One of the approaches used to forecast future events
worldwide is the trend analysis. Trend analysis is based
on the assumption that the contributing factors will still
be dominant at the point of estimation and will stay
valid. Thus, the next future events will be easily
expected by simple extrapolation. However, this may
be done with caution. Even as the trend in a region
over years has been one of fewer fatalities, it is quite
possible that with economic development and rapid
motorization there is potential for growth in the number
of accidents and fatalities.
Page | 207

Global Journal of Advanced Engineering Technologies


ISSN (Online): 2277-6370 & ISSN (Print):2394-0921
Table 1 shows the changes in number of total accidents,
injuries, and fatalities in Kuwait during the period
1997-2014. The changes in these safety parameters over
time are depicted in figure 1. It can be seen that during
the 18 year period, there has been an increase of about
400% in accidents accompanied by about 50% increase
in fatalities. This rapidly increasing rate alerts
responsible authorities to take all necessary actions to
face this serious problem. Road traffic fatalities are
increasing but at a lower rate than accidents. The
reduction in the number of injuries over the same
period especially during the recent years is mainly due
to a better documentation and coordination between
General Traffic Department and Hospitals (medical
Emergency). Fatalities have increased with time until
they reached their maximum in 2011, and then
declined. This can be explained by the expanding of
using the smart-phones while driving in 2011. Drives
became more aware of the effect of this behaviour on
their safety which, together with increased police
surveillance, resulted in reduction of casualties.
TABLE 1: Changes in number of accidents, injuries,
and fatalities (1997-2014)
Total
Number of Number of
YEAR
Number Of
Injuries
Fatalities
Accidents
1997
26426
1681
360
1998
27456
1690
334
1999
26635
1743
333
2000
27696
1125
331
2001
31028
1566
300
2002
37650
2249
315
2003
45376
1332
372
2004
54878
824
398
2005
56235
863
451
2006
60410
853
460
2007
63323
1014
447
2008
56660
1095
410
2009
61298
670
407
2010
65861
786
374
2011
75194
971
493
2012
85557
9959
454
2013
89527
8977
445
2014
99047
8783
461
Many accident rates are used to compare the safety
performance of different locations and to prioritize
safety improvements. These include the rates that
relate the number of accidents, injuries or fatalities to
safety parameters such as population, number of
registered vehicles, and distances travelled expressed in
vehicle-km.

www.gjaet.com

Volume 4, Issue 3- 2015


120000
100000
80000

Accident
injury
fatalities

60000
40000
20000
0

Figure 1: Trends of Total Accidents, Injuries and


Fatalities
TABLE 2: Changes in fatality rates during the period
(1997-2014)
Vehicle
Fats/100,000 Fats/10,000
YEAR
Ownership
persons
Vehicles
18.1846744
1997
0.330691841 6
5.49897887
16.4767157
1998
0.34970152
9
4.71165118
15.8029987
4.41351888
1999
0.358058936 7
7
15.1164468
4.31097919
2000
0.350649962 8
1
13.1869291
3.42614376
2001
0.384891296 2
1
13.0650748
3.32495234
2002
0.392940214 6
2
14.6072304
3.89537769
2003
0.374988809 2
5
14.4535119
3.81731738
2004
0.378630083 9
5
15.0776162
3.97692501
2005
0.379127497 9
7
14.4519566
3.58823589
2006
0.402759381 7
1
13.3035714
3.27522915
2007
0.4061875
3
6
11.8840579
2.86986224
2008
0.414098551 7
7
11.6285714
2.72139774
2009
0.427301429 3
2
10.4866565
2.39581104
2010
0.437707998 7
6
13.3340834
3.00570352
2011
0.443626038 3
1
11.8732294
2.76102982
2012
0.430029019 8
8
11.2231407
2.54514923
2013
0.440961992 5
2
11.2659039
2.50907976
2014
0.449005411 3
6
Page | 208

Global Journal of Advanced Engineering Technologies


ISSN (Online): 2277-6370 & ISSN (Print):2394-0921
Table 2 shows the changes in vehicle ownership and
fatality rates in Kuwait during the period (1997-2014).
It can be seen that during the 18-year period, there has
been an increase of almost 37 percent in vehicle
ownership accompanied by 87 percent reduction in
fatality rate (per 10,000 vehicles) as compared to the
growth of population were 45 percent reduction in
fatality rate per 100,000 persons.
III. MODEL DEVELOPMENT
This section describes the process of developing the
road fatality prediction model for Kuwait. As fatalities
are fully reported and therefore confidently used in the
model development, a process which involved two
primary steps; the first is the selection of the variables
that are believed to have significant effect on fatalities
and for which sufficient reliable data are available. The
second step involved the creation of a representative
model that incorporates the selected variables.
A) Selection of Model Variables
The key data necessary to build the model were drawn
from the Kuwaiti General Traffic Department, Public
Authority for Civil Information, and Ministry of
planning (Annual Statistical Abstracts) [10] for the
years 1997 through 2014. The independent variables
that satisfy the selection criteria were population,
number of registered vehicle, and the length of paved
roads. Table 3 presents the time-series data of these
variables for the designated time period.
TABLE 3: Road fatalities, population, registered
vehicles, and total length of paved roads (1997 2014)
Total
Number of length
Year Fatalities
Population Registered
of paved
Vehicles
roads
(km)
1997 360
1979689
654667
4954
1998 334
2027103
708881
4967
1999 333
2107195
754500
5213
2000 331
2189668
767807
5230
2001 300
2274980
875620
5283
2002 315
2411008
947382
5445
2003 372
2546684
954978
5536
2004 398
2753656
1042617
5749
2005 451
2991189
1134042
5794
2006 460
3182960
1281967
5959
2007 447
3360000
1364790
6101
2008 410
3450000
1428640
6343
2009 407
3500000
1495555
6525
2010 374
3566437
1561058
6609
2011 493
3697292
1640215
6996
2012 454
3823728
1644314
7180
2013 445
3965022
1748424
7321
2014 461
4091993
1837327
7516
www.gjaet.com

Volume 4, Issue 3- 2015

B) Model Development and Results


The road fatality model was created using Multiple
Linear Regression Modelling (MLRM). The analysis
was carried out using the Statistical Package for Social
Sciences Version 20 (SPSS 20.0). A multiple regression
equation for predicting Y can be expressed a follows:
Y = A + B1X1+ B2X2+ B3X3 +..+BiXi +
(1)
Where: Y is the predicted value of the dependent
variable,
A is the constant coefficient of the regression line or
intercept,
Xi is the dependent variable,
Bi is the regression coefficient of Xi.
The performance of the model is evaluated based on
the R-squared (R2) value of the regression line. Its
value lies between 0 & 1. The higher the R-squared
value the better the performance of the model. The data
was collected from the Kuwaiti General Traffic
Department, Public Authority for Civil Information,
and Ministry of planning for the 18 consecutive years
starting from the year 1997 as shown in Table 3.
In this regression model there are three predictor
variables, namely population (P), number of registered
vehicles (V) and total Length of paved roads (L). The
dependent variable was the number of fatalities (F).
Descriptive statistics of these variables are shown in
Table 4.
The regression equation with all three predictor
variables was significantly related to fatalities with a
coefficient of determination R2 = 0.803, and adjusted R2
= 0.761and the model coefficients are shown in Table
5and Table 6. The final regression model developed
using the available data was found to be of the
following form:
F = -140.358 + 4.636P -4.227V + .404L
(2)
Where: F is the predicted number of fatalities,
P is the population,
V is the number of registered vehicles,
L is the total length of paved roads (km).
TABLE 4: Descriptive Statistics
Standard
Variables
Mean
N
Deviation
Fatalities
396.94
59.209
18
Population 2995478.00 719666.716 18
Registered
1213488.00 389629.628 18
Vehicles
Length of
Road
in 6040.06
829.287
18
(KM)
TABLE 5: Model Summary
Standard
Adjusted
R
R Square
Error of
R Square
Estimate
.896
.803
.761
28.975
Page | 209

Global Journal of Advanced Engineering Technologies


ISSN (Online): 2277-6370 & ISSN (Print):2394-0921

Volume 4, Issue 3- 2015

TABLE 6: Model Coefficients


Un standardized Coefficients

Model
(Constant)
Population
Registered Vehicles
Length of paved
Roads

B
-140.358
.000
-.001

Std. Error
212.904
.000
.000

.029

.051

There are various indices that show the strength of


relationship between dependent variable and various
predictor variables. It shows as how we can present the
dependent variables in the regression model by using
various predictor variables. The value of R Square (.803)
indicates that 80.3% of the fatalities can be accounted for
by its linear relationship with the three predictor
variables implying that these three predictor variables
can represent a good regression model for predicting
fatalities in Kuwait.
TABLE 7: The Bivariate and Partial Correlations of the
Predictors with Fatality Index

Predictors

Correlation
between each
Predictor and
the
Fatality
index

Correlation between
each Predictor and the
Fatality
index
controlling for all
other Predictors

Population

.823**

.696**

.787**

-.576*

.769**

.150

Registered
Vehicles
Length of
Road

Table 7 presents the relative strength of the individual


predictors. As expected, all the bivariate correlations
between predictors and fatality variable were positive
and statistically significant at (.01). Two out of three
partial correlations were statistically significant. Partial
correlation between population and fatality and between
registered vehicle and fatality were significant at (.01).
On the basis of this correlations analysis, it is concluded
that only two out of three predictors (population and
registered vehicles) were useful in significantly
predicting the fatality. It is interesting to note that
predictor variable population alone accounted for
68% (.823*.823 = .677) of the variance of the fatality
index and the predictor variable registered vehicle
alone accounted for 62% (.787*.787 = .619) of the
variance of the fatality index. However the judgments
about the relative importance of these predictors are
difficult because they are significantly correlated with
each other as shown below.
www.gjaet.com

Standardized
Coefficients
Beta

Sig

4.636
-4.227

-.659
3.603
-2.636

.404

.568

.520
.003
.020
.579

The correlation between population and registered


vehicle is r (16) =.995, p<.001, the correlation
between population and length of road is r (16)
=.977, p<.001 and the correlation between length of
road and registered vehicle is r (16) =.985, p<.001.
IV. A RECOMMENDED NATIONAL STRATEGY
Accidents cannot be eliminated totally but a lot can be
done to effectively reduce their magnitude and severity.
After deciding upon the factors which contributed to
accidents occurrence, it is necessary to identify some
potential countermeasures to improve road safety in
Kuwait. This requires the adaptation of two strategies
namely accident reduction and accident prevention.
Accident prevention can be achieved through better
planning and design of new roads. Considerable
freedom exists when planning and designing new roads
and it may be possible to introduce significant
improvements in road safety terms at minimal cost if
such issues are given sufficient attention at an early
stage
The first group of potential countermeasures includes
such measures as: Road safety education at schools,
road user training, road safety publicity, improved
driver and vehicle licensing and testing, stricter law
enforcement, improved pedestrian facilities, road
design and maintenance measures (safety audits at all
stages of road construction and maintenance), traffic
engineering, control and safety measures (traffic
calming, work zone safety management),in addition to
improvement of trauma care and emergency services
associated with the availability of prompt and expert
medical attention. Besides the traditionally successful
countermeasures suitable under Kuwaiti driving
conditions, a strategic National Road Safety Plan
(NRSP) is considered necessary to encompass the
various negative impacts of road accidents in Kuwait.
The proposed NRSP is to be implemented through a
national consultative process with defined measureable
indicators and setting targets which are formally
endorsed by the government and supported by specific
budgetary allocations. The purpose of the NRSP is to
establish a coordinated approach which will assist in
devising the most appropriate solutions for the road
safety problems; implementing these solutions; and
evaluating them to improve future program design.
Page | 210

Global Journal of Advanced Engineering Technologies


ISSN (Online): 2277-6370 & ISSN (Print):2394-0921
The main features of the plan include:
Coordination of resources, activities and ideas
in Kuwait across all sectors.
A summary of road safety programs scheduled
to be undertaken at both national and local
levels; and a planning process to facilitate the
collective and individual targeting of
resources by a range of agencies into road
safety programs.
The definition of target areas for road safety
programs to achieve the best results for the
least cost.
Annual revision of both the plan and targets in
order to create appropriate set of guidelines
for the situation as it develops.
The creation of a descriptive database of all
the existing programs freely accessible to any
department or organization with an interest in
road safety.
The targets of the plan represent the problem areas in
road safety which must be addressed in the process to
improve safety performance. The target areas include
people and
operators, roads, vehicles and systems
(training and emergency services). It is considered that
the above NRSP can be achieved successfully through
sufficient allocated resources, and coordinated effort
and commitment among safety professionals in
engineering, enforcement, education, and public
officials at both national and local level.
V. CONCLUSION
Fatalities are fully reported and were therefore
confidently used for the development of a road fatality
prediction model as described in this work .The model
can be used to estimate road fatalities in Kuwait as a
function of population, number of registered vehicles
and length of paved roads. The model was found to be
statistically significant with a coefficient of
determination (R2) of 0.803 meaning that the three
contributory factors explain 80.3% of the variability in
fatalities. Although the developed model has a good
explanatory power, it is recommended that further
research effort should be given to introduce other
factors affecting road accidents fatalities, with a greater
levels of detail than what is generally available now.
The results of accident trend analysis in Kuwait reveal
the need to identify some potential countermeasures to
improve road safety. This requires the adaptation of two
strategies namely accident reduction and accident
prevention. It is realized that the eventual effectiveness
of many road safety initiatives in Kuwait is blundered
by the fact that a diversity of well-intentioned
organizations and individuals are involved with road
safety and nobody is admitting responsibility for
accident and causality reduction. Therefore, in order to
effectively encompass the problem, a recommended
National Road Safety Plan as detailed in the paper is
considered necessary.
www.gjaet.com

Volume 4, Issue 3- 2015

ACKNOWLEDGMENT
The Author would like to thank the Kuwaiti General
Traffic Department, Public Authority for Civil
Information and Ministry of planning for providing the
necessary data.
REFERENCES
[1] Al-Matawah, J and Jadaan, K. S. 2009, Application
of Prediction
Techniques to Road Safety in
Developing Countries, International Journal of Applied
Science
and Engineering. Vol. 7, 2: 169-175.
[2] Arndt, O. K., Troutbeck, R. J., 1998,Relationship
between round about geometry and accident rates
28:1-16.Available,from
www.onlinepubs.trb.org
/onlinepubs/circulars/ec003/ch28.pdf), [Accessed 20
December, 2010].
[3]Baek, S., Chang, H., Kang, J., Yoon, B., 2005, An
effect of trip length on freeway accident rates Journal
of the Eastern Asia Society for Transportation Studies,
Vol. 6, pp. 3467 - 3481.
[4] Bener, A., Hussain, S., Al-Malki, Shotor, M.
andJadaan, K. 2010. Road Traffic Fatalities in Qatar,
Jordan and the UAE: Estimates Using Regression
Analysis and the Relationship With Economic Growth.
Eastern Mediteranean Health Journal (EMHJ),
15(3):318-323.
[5] Hadayeghi, Shalaby, and Persaud, 2003, Macrolevel Accident Prediction Models for Evaluating the
Safety
of
Urban
Transportation
systems,
Transportation Research Record 1840, 87-95
[6] Jadaan, K. S., Khalil, R. and Bener, A., 1991, A
Mathematical Model using Convex Combination for the
prediction of Road Traffic deaths. Computing
andInformation, 1991, 2:139-157.
[7]Jadaan, K. S., Alenezi, F., AlZahrani, 1992, A.
Road fatalities in Saudi Arabia: Trends and
prediction,Proceeding of the REAA workshop Vol.3;
19-24.
[8] Jadaan. k., Al-Fayyad, M., and Gammoh, H.
2014.Prediction of Road Traffic Accidents in Jordan
using Artificial Neural Network. Journal of Traffic and
Logistics Engineering, 2: 92-94.
[9] Marchionna A., Perco P., Tavernar M.C.,2008,
Transferability of accident prediction models for urba
intersections, Proceedings of the 17th National SIIV
Congress,
Italy,
2008.
Available
from:
(www.siiv.it/documenti/63_2848_20080919095714.pdf)
, [Accessed 17 December, 2010].
[10]Ministry of Planning, 2014, Annual Statistics
Abstracts, Kuwait.
[11] Mohamed, O.A.I., 2008, An Econometric Model
for Road Crashes in Khartoum State, Ph.D. Thesis,
Faculty of Economic and Social Studies, University of
Khartoum, Sudan.
[12] Nambuusi, B.B., Brijs, T., Hermans, E.,2008, A
review of accident
prediction models for road
intersections, Steunpunt Mobiliteit & Open bare
Werken , 2008, p.56-57.
Page | 211

Potrebbero piacerti anche