Research on the Influence of Lane Width on Drivers' Dangerous

Lane Change Behavior on Urban Roads

Yuting Gao

School of Transportation Engineering, Chang’an University, Xi’an, Shannxi, 710000, China

Keywords: Lane Width, Drivers' Dangerous Lane Change Behavior, Linear Regression.

Abstract: Dangerous lane change behavior by drivers (DLCB) is defined as the behavior of the driver manipulating the

vehicle to produce excessive lane change behavior, which causes a certain degree of danger to the surrounding

vehicles or pedestrians. This paper investigates whether lane width and other factors have an impact on DLCB,

and explores the correlation between them. Multiple linear regression was used to analyze the relationship

between the number of dangerous lane changes and lane width, traffic flow, climate, light intensity and time.

Based on the results of the fitting, it can be found that lane width and traffic volume can explain more than

73% of the changes in DLCB. In addition, the reason for the analysis error may be that the bus stop does not

have a bus lane, which causes the bus behind to be forced to change lanes when the bus stops at the stop. At

present, there is no specific relationship between lane width and DLCB, but it is only pointed out that there

may be a correlation between the two. Therefore, the research in this paper can provide ideas for future road

design and planning to a certain extent.

1 INTRODUCTION

Lane width generally refers to the width of the lane

drawn on the road surface by markings, and the width

is determined according to the vehicle type, vehicle

speed, vehicle operation characteristics, etc. For a

long time, it was widely believed that the wider the

road, the safer it was (Zhu, 2023). The current road

engineering design code in China uses the vehicle as

the basic reference to determine the allocation of road

space, resulting in a clear preference for motor

vehicles in the right of way (Jiang 2022). Therefore,

the lanes are generally designed to be wider.

Theoretically, as the width of the lane decreases, the

road capacity decreases. However, in practice, there

is a high probability of uncivilized driving behaviors

(arbitrary lane changes, speeding, irregular

overtaking, etc.) in wide lanes, so a reasonable

reduction and adjustment of lane widths can limit

driving behavior and increase road capacity (Su

2022).

Lane change behavior is a driving behavior that

occurs more often in road traffic. According to

statistics, traffic accidents caused by inappropriate

lane changes account for about 10% of the total

number of accidents (Ning 2023). In addition,

frequent lane changes in congested road sections will

affect the overall traffic efficiency of traffic flow to a

certain extent, causing traffic congestion (China

Journal of Highway and Transport 2017). At present,

the most widely used types of lane change are divided

into mandatory lane change and arbitrary lane change

(Qi 2020). Lane change behavior is related to the

driver's hazard perception. Of all the skills related to

driving, only hazard perception is closely related to

traffic accidents (Wang 2019). Due to the driver's

incomplete observation of the traffic environment,

the distance between the vehicles that can be inserted

in the target lane is small, which may cause the

vehicle in the lane change to collide with the vehicle

in front or behind the target lane (Xia 2022).

Dangerous lane change can be defined as the behavior

of the driver manipulating the vehicle to change lanes

excessively, causing a certain degree of danger to

surrounding vehicles or pedestrians. If the lane

change time is defined by the lane change time, a lane

change time of less than 3 s is considered a dangerous

lane change (Qi 2020). Dangerous lane changes can

be subdivided into frequent lane changes, S-shaped

lane changes, continuous lane changes, and too fast

lane changes.

This paper focuses on arbitrary lane change

behavior and the influence of lane width on the

Gao, Y.

Research on the Inﬂuence of Lane Width on Drivers’ Dangerous Lane Change Behavior on Urban Roads.

DOI: 10.5220/0012797500004547

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Data Science and Engineering (ICDSE 2024), pages 405-409

ISBN: 978-989-758-690-3

405

number of dangerous lane changes of drivers is

studied, and the factors such as traffic flow, climate,

and light intensity are comprehensively considered.

METHOD

2.1 Data Sources and Descriptions

Combined with the street type and specific

construction conditions, the minimum width of a

motor lane in the road section can be narrowed to

2.85~3.25m according to different design speeds

(Jiang 2022). In this paper, three urban roads with a

width of 3.1 meters, 3.32 meters, and 3.45 meters

were selected as the research objects, and a total of 25

minutes of traffic flow videos were taken in the

morning (8 a.m. to 10 a.m.), noon (12 p.m. to 2p.m,),

and evening (18 a.m. to 8:00 p.m.) during the week

and on weekends, respectively, according to the ideas

shown in Fig. 1. At the same time, the climate of the

day is recorded.

2.2 Indicator Selection and Description

In this paper, the relationship between the number of

dangerous lane changes and lane width, light

intensity, traffic volume, and climatic conditions was

studied every 5 minutes. Traffic volume is defined in

this article as the total number of vehicles passing

through during that period. In addition, considering

that the number of lanes is different, it is difficult to

compare the number of lane changes, so the three

roads selected are all three lanes. Because the impact

of buses, cars and motorcycles on the road is

different, according to the vehicle conversion factor,

a bus can be converted into two cars, and a car can be

converted into two motorcycles, to convert the

number of vehicles and the number of times of the

driver's dangerous lane change behavior. Table 1

shows a portion of the DLCB of climate, temperature,

light intensity, and traffic volume recorded at the time

of data collection, as well as the measured lane width.

2.3

Method

Fig. 2 and Fig. 3 plot the relationship between the lane

width and the traffic volume of the DLCB,

respectively, and it can be seen that there is a linear

relationship between the lane width and the traffic

volume and the DLCB. Therefore, this paper is more

suitable for multiple linear regression analysis to

analyze the numbers.

Figure 1: Data collection ideas (Original).

Table 1: Some data are displayed.

Numbering Lane width/m Time Weather Temp. Light Num. of vehicles DLCB

1 3.1 12:35 cloud

20 strong 113 3

2 3.32 9:50 Sun. 7 Weake

164.5 6.5

3 3.45 12:40 cloud

20 strong 216 7

ICDSE 2024 - International Conference on Data Science and Engineering

406

Figure 2: Scatterplot of the lane width and DLCB (Original).

Figure 3: Scatterplot of the number of traffic and DLCB (Original).

Linear regression is a simple method of

regression analysis that explores the linear

relationship between two variables. Multiple linear

regression is an extension of linear regression that

explores the linear relationship between multiple

independent variables and one dependent variable

(Zhang 2022). The principle is to estimate the

regression coefficient by minimizing the sum of

squares of the residuals, so that the residuals between

the predicted and observed values are minimized. It

can help to understand the relationship between

variables and can help us determine the effect of

different independent variables on the dependent

variable, and in turn, determine the most relevant

independent variables (Jiang 2022). At the same time,

it provides a hypothesis test on whether the regression

coefficient is significant to help evaluate the

statistical significance of the model. Therefore, it is

appropriate to choose multiple linear regression in

this paper.

This paper examines whether the number of

dangerous lane changes is related to lane width.

Therefore, linear regression can be used to study

whether there is a correlation between the two, that is,

the number of dangerous lane changes of drivers is

the dependent variable Y, and the lane width is the

independent variable X. At the same time, it is also

necessary to consider the time (there may be

differences in the state of the driver in the morning,

noon and evening), climate environment, light

intensity and other factors, so the data processing

method can be used by multiple linear regression. In

addition, in order to test the rationality of the model,

20% of the data were randomly selected for the test

of the final regression equation, and the remaining

80% was used for regression training.

3 RESULT AND DISCUSSION

3.1 Data Analysis

The results of linear regression fitting are shown in

Table 2 with lane width, traffic flow, weather,

temperature and light intensity as the independent

variables, and











the number of

dangerous lane changes of drivers as the dependent

variables Y.

Research on the Inﬂuence of Lane Width on Drivers’ Dangerous Lane Change Behavior on Urban Roads

407

Table 2: Fitting results of traffic volume, lane width, climate, etc., to the number of DLCB.

B standard error Beta t p VIF Tolerance

constant -7.638 2.476 - -3.085 0.003 - -

Lane width 1.990 0.837 0.149 2.376 0.019 1.836 0.545

Traffic Num. 0.036 0.003 0.750 11.846 0.000 1.872 0.534

Weathe

-0.116 0.202 -0.027 -0.576 0.566 1.050 0.952

temperature 0.003 0.015 0.011 0.231 0.818 1.091 0.917

ht intensit

0.122 0.098 0.059 1.245 0.216 1.061 0.943

0.743

ust R

0.732

F F (5,120) =69.408, p=0.000

D-W values 1.687

Dependent variable： Number of dangerous lane change behavior

* p<0.05 ** p<0.01

Table 3: Linear regression analysis results.

B standard erro

Beta t

VIF Tolerance

constant -7.173 2.413 - -2.973 0.004 - -

Lane width 1.870 0.818 0.140 2.287 0.024 1.766 0.566

Traffic 0.036 0.003 0.761 12.427 0.000 1.766 0.566

0.739

Adjust R

0.735

F F (2,123) =174.002, p=0.000

D-W values 1.657

Dependent variable： Number of dangerous lane change behavior

* p<0.05 ** p<0.01

As can be seen from the above table 2, the model

formula is as follows:

𝑌 = −10.660 + 2.893𝑋



+ 0.036𝑋



− 0.077𝑋



(1)

The R 2 value of the model is 0.743, which means

that lane width, traffic flow, weather, temperature,

light intensity, and time can explain 74.3% of the

change in the Number of dangerous lane change

behavior. When the model was tested F, it was found

that the model passed the F test, which means that at

least one of the lane width, traffic flow, weather,

temperature, light intensity, and time would affect the

number of dangerous lane change behaviors. In

addition, the multicollinearity of the model is tested,

and it is found that all the VIF values in the model are

less than 5, which means that there is no collinearity

problem, and the D-W value is around the number 2,

which indicates that there is no autocorrelation

between the model and the sample data. According to

the summary analysis of the t-test and the p-test, it can

be found that the number of traffic, the lane width has

a significant positive effect on the number of

dangerous lane change behavior. However, weather,

temperature, light intensity, and time of day do not

affect the number of dangerous lane change

behaviors.

Excluding the factors of weather, temperature,

light intensity, and time, the results of linear

regression fitting are shown in Table 3.

As can be seen from the table 3 above, the model

formula is:

Y = -7.173 + 1.870X



+ 0.036X



(2)

The model R 2 value is 0.893, which means the

lane width, and the traffic volume can explain 89.3%

of the change in the number of dangerous lane change

behaviors.

3.2 Discussion

According to the results of multiple linear regression

fitting, it can be seen that there is a linear regression

relationship between the number of dangerous lane

change behaviors and lane width and traffic flow, but

the fitting results can only explain 89.3% of the

change reasons. This may be because there are bus

stops on the selected section, but there are no bus

lanes, so when the bus stops, the cars behind can only

change lanes. However, the frequency, stop time, and

number of stops on the three roads are not the same,

so the influence of bus stops on the number of

dangerous lane changes cannot be ruled out.

Although it is not possible to rule out the influence of

bus stops on the number of dangerous lane changes,

it is clear that the number of dangerous lane changes

of drivers is indeed related to lane width.

ICDSE 2024 - International Conference on Data Science and Engineering

408

4 CONCLUSION

Dangerous lane change refers to a driver's behavior of

excessively changing lanes, posing a certain degree

of danger to surrounding vehicles or pedestrians. In

this paper, three urban roads with different lane

widths in Yan'an, Shaanxi Province, were selected to

collect data on traffic flow, climate, and the number

of dangerous switching behaviors in 5 minutes.

Linear regression analysis was carried out with lane

width, traffic flow, weather, temperature, light

intensity, and time as independent variables, and the

number of dangerous lane change behaviors of

drivers as dependent variables. According to the final

results, it can be found that the traffic flow of lane

width significantly affects the number of dangerous

lane change behaviors, and can explain 89.3% of the

change of dangerous lane change behaviors.

However, the bus stop does not set up a bus lane,

which will have a significant impact on the driver to

change lanes, so it will lead to a large part of the data

test.

REFERENCES

R. Zhu, Experimental study on the measurement of driver's

psychological stress under lane constraints,

(2023)https://link.cnki.net/doi/10.27242/d.cnki.gnjlu.2

023.000100.

Y. Jiang, Urban Planning, 46,62-70, 2022.

X. Su, Journal of Wuhan University of Technology

(Transportation Science and Engineering), 46, 592-597,

2022.

H. Ning, Research on the relationship between unsafe

driving behavior and road traffic accidents, (2023)

https://link.cnki.net/doi/10.27205/d.cnki.gltec.2023.00

0010.

China Journal of Highway and Transport, 30, 1-197(2017).

Y. Qi, Wuhan University of Technology, 05, 85(2020).

M. Wang, Transportation Science and Engineering, 35,

106-112(2019).

H. Xia, Research on the identification method of dangerous

driving behavior in the merging area of urban

expressway based on vehicle trajectory,

(2022)https://link.cnki.net/doi/10.27671/d.cnki.gcjtc.2

022.000536.

H. Zhang, Automation and Instrumentation, 1-4+8, (2022).

J. Jiang, Research on heteroskedasticity in multiple linear

regression models, (2022) https://link.cnki.net/

doi/10.27050/d.cnki.gglgc.2022.000.

Research on the Inﬂuence of Lane Width on Drivers’ Dangerous Lane Change Behavior on Urban Roads

409