ArticlePDF Available

Behavior Patterns of Long-term Car-sharing Users in China*

Authors:

Abstract and Figures

This paper presents the behavior patterns of long-term users in detail on the basis of empirical data of car sharing in Hangzhou, China. Users, whose utility time are more than three months and frequency of usage are beyond once per month, have been selected as the subject investigated (long-term users) in this study. Unlike other studies on car sharing in China, which only conducted theoretical analyses and investigation of willingness for car sharing, this study was based on real operation data of Hangzhou, which was the first massive pilot city for car sharing project in China. The major objective of this study was to analyze the differences between groups classified from the long-term users group according to the different frequency of usage, and identify and summarize the typical usage patterns by using indices such as new members, monthly orders, single-use time, single-use distance, time of taking-out and placing-in a car, and so on. The findings indicate that the behaviour patterns of each group are different: that of the highest frequency users group are similar to the characteristics of commuting travel and that of lowest frequency users group are more similar to the car-sharing users abroad. The key contribution of this paper is presenting the different behaviour patterns of the Chinese users in groups differing in frequency, and act as a foundation for questionnaire surveys and policy analysis in the future.
Content may be subject to copyright.
ScienceDirect
Available online at www.sciencedirect.com
Transportation Research Procedia 25C (2017) 4666–4682
2352-1465 © 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
10.1016/j.trpro.2017.05.303
www.elsevier.com/locate/procedia
10.1016/j.trpro.2017.05.303
© 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
2352-1465
Available online at www.sciencedirect.com
ScienceDirect
Transportation Research Procedia 00 (2017) 000–000
www.elsevier.com/locate/procedia
2214-241X © 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
World Conference on Transport Research - WCTR 2016 Shanghai. 10-15 July 2016
Behavior Patterns of Long-term Car-sharing Users in China*
Ying HUIa* Wei WANGa Mengtao Dinga Yian Liub
a:Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji University
b:Key Laboratory of Chelizi Intelligent Technology Company
Abstract
This paper presents the behavior patterns of long-term users in detail on the basis of empirical data of car sharing in
Hangzhou, China. Users, whose utility time are more than three months and frequency of usage are beyond once per
month, have been selected as the subject investigated (long-term users) in this study. Unlike other studies on car
sharing in China, which only conducted theoretical analyses and investigation of willingness for car sharing, this
study was based on real operation data of Hangzhou, which was the first massive pilot city for car sharing project in
China. The major objective of this study was to analyze the differences between groups classified from the long-
term users group according to the different frequency of usage, and identify and summarize the typical usage
patterns by using indices such as new members, monthly orders, single-use time, single-use distance, time of taking-
out and placing-in a car, and so on. The findings indicate that the behaviour patterns of each group are different: that
of the highest frequency users group are similar to the characteristics of commuting travel and that of lowest
frequency users group are more similar to the car-sharing users abroad. The key contribution of this paper is
presenting the different behaviour patterns of the Chinese users in groups differing in frequency, and act as a
foundation for questionnaire surveys and policy analysis in the future.
© 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
Keywords: car-sharing; long-term users; behavior patterns; group
* National Natural Science Foundation of China, No.51408430
* Corresponding author. Tel:.13636458669
E-mail address: huiying@tongji.edu.cn
Available online at www.sciencedirect.com
ScienceDirect
Transportation Research Procedia 00 (2017) 000–000
www.elsevier.com/locate/procedia
2214-241X © 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
World Conference on Transport Research - WCTR 2016 Shanghai. 10-15 July 2016
Behavior Patterns of Long-term Car-sharing Users in China*
Ying HUIa* Wei WANGa Mengtao Dinga Yian Liub
a:Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji University
b:Key Laboratory of Chelizi Intelligent Technology Company
Abstract
This paper presents the behavior patterns of long-term users in detail on the basis of empirical data of car sharing in
Hangzhou, China. Users, whose utility time are more than three months and frequency of usage are beyond once per
month, have been selected as the subject investigated (long-term users) in this study. Unlike other studies on car
sharing in China, which only conducted theoretical analyses and investigation of willingness for car sharing, this
study was based on real operation data of Hangzhou, which was the first massive pilot city for car sharing project in
China. The major objective of this study was to analyze the differences between groups classified from the long-
term users group according to the different frequency of usage, and identify and summarize the typical usage
patterns by using indices such as new members, monthly orders, single-use time, single-use distance, time of taking-
out and placing-in a car, and so on. The findings indicate that the behaviour patterns of each group are different: that
of the highest frequency users group are similar to the characteristics of commuting travel and that of lowest
frequency users group are more similar to the car-sharing users abroad. The key contribution of this paper is
presenting the different behaviour patterns of the Chinese users in groups differing in frequency, and act as a
foundation for questionnaire surveys and policy analysis in the future.
© 2017 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of WORLD CONFERENCE ON TRANSPORT RESEARCH SOCIETY.
Keywords: car-sharing; long-term users; behavior patterns; group
* National Natural Science Foundation of China, No.51408430
* Corresponding author. Tel:.13636458669
E-mail address: huiying@tongji.edu.cn
2 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
1.Introduction
Against the background of rapid urbanization and the rapid rise in private car ownership in China, balancing
resources, efficiency, and fairness while meeting the diverse travel needs of residents has become an extremely
difficult challenge. Car sharing, which offers the mobility and flexibility of private cars and at the same time eases the
pressure on public transportation can reduce private car ownership to a certain extent (Meijkamp, 1998). This is being
increasingly considered as an innovative mobility tool from the viewpoint of transportation policy in Europe and other
developed countries (Ohta et al, 2009).
The practice of car sharing dates back to the 1990s in European countries. Shaheen classified the progress of car
sharing into three phases: initial market entry and experimentation phase (19942002); growth and market
diversification phase (20022007); and commercial mainstream phase (2007present). (Shaheen, 2009) It has already
reached the third phase in some of the developed countries in North America. Car sharing has become not only a part
of numerous urban transportation systems, but also an effective means of reducing car ownership and the total mileage
(Stillwater et al, 2008). Most initial studies on car sharing in North America dealt with the mobility and flexibility of
the car-sharing system, its influence on the urban transportation system, and ways to expand car-sharing membership;
however, the behavior patterns of car-sharing users was rarely explored in these studies. However, researchers soon
discovered that the characteristics of user behavior were significant to topics studied in the previous works because
behavior patterns may explain some system operation problems from the root itself. Several researchers, for example,
Nobis (2006), Millard-Ball (2006), Celsor and Millard-Ball (2007), Morency (2008, 2009), and Khandlker M. et al
(2011) analyzed usersbehavior patterns, and they all considered it as an important factor determining the success of
the car-sharing system operation.
Thus far, several studies have been conducted outside China on the behavior patterns of car-sharing users. Seik
(2000) analyzed actual car-sharing conditions, investigated the characteristics of travel distance and travel destination
in Singapore, and discovered significant differences between members and non-members. Some studies (2009, 2010)
showed that after joining a car-sharing club, members mainly used car-sharing systems for long holidays or for
shopping and travel; for regular commuting, the members always used public transportation. In the terms of the
frequency and distance of usage, some studies in England showed that 75% of the members used car-sharing less than
5 times per year, that the distance travelled thus by 64% of the members was less than 40 miles. Some studies in
Belgium indicated that used car-sharing less than thrice a month and mainly used it for shopping, visiting friends, and
taking holiday in their leisure time. Millard-Ball (2005) also found that only a small proportion of the members used
car-sharing for regular commuting, and most members used car-sharing for carrying heavy things or for trips with
many destinations. Khandlker et al. (2011) analyzed the empirical data and found that men and French-speaking
members tend to practice car sharing only for short trips, but their usage frequency is high. Meanwhile, it could not be
proved that increasing the number of cars has any effect on maintaining the membership, but it may lead to an increase
in the frequency of usage to some extent.
Unlike in North America, in China, car sharing is only in the initial stage. Because of the lack of practical support,
relatively less research has been carried out on car sharing in China. Huang Zhaoyi (2000) summed up the car-sharing
concept and proposed a number of car-sharing initiatives. Many researchers summarized and analyzed the practice
abroad. Hui Ying (2008) and Ye Liang (2012) opined that by actively encouraging car sharing, households can be
guided into using this alternative rather than buying a car, thus slowing down the trend of households buying a second
car. After reviewing car-sharing practices in cities abroad, Xu Qing (2014) reported various problems related to car
sharing in China. Qie Lisha studied the empty car distribution and created a model to validate this distribution for car-
sharing networks. Although some studies on the behaviors of car-sharing users are noteworthy, most still dealt only
with the level of willingness. Hui Ying (2010, 2012, 2013) and Wang Zengquan (2013) used discrete choice models
for the willingness of joining a car-sharing club and analyzed the potential requirements and possibility of changing
travel behavior.
In China, car-sharing studies just stayed in the level of theoretical analysis and willingness investigation, extremely
lack of the study on members' behavior patterns, which was based on the empirical data. Based on the no-personal
privacy information car rental data of a car-sharing system initiated in July 2013 in Hangzhou, China, this work
analyzed the behavior patterns of long-term users. These users were classified into several groups, and the typical
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4667
Available online at www.sciencedirect.com
ScienceDirect
Transportation Research Procedia 00 (2017) 000–000
www.elsevier.com/locate/procedia
Available online at www.sciencedirect.com
ScienceDirect
Transportation Research Procedia 00 (2017) 000–000
www.elsevier.com/locate/procedia
2 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
1.Introduction
Against the background of rapid urbanization and the rapid rise in private car ownership in China, balancing
resources, efficiency, and fairness while meeting the diverse travel needs of residents has become an extremely
difficult challenge. Car sharing, which offers the mobility and flexibility of private cars and at the same time eases the
pressure on public transportation can reduce private car ownership to a certain extent (Meijkamp, 1998). This is being
increasingly considered as an innovative mobility tool from the viewpoint of transportation policy in Europe and other
developed countries (Ohta et al, 2009).
The practice of car sharing dates back to the 1990s in European countries. Shaheen classified the progress of car
sharing into three phases: initial market entry and experimentation phase (19942002); growth and market
diversification phase (2002–2007); and commercial mainstream phase (2007–present). (Shaheen, 2009) It has already
reached the third phase in some of the developed countries in North America. Car sharing has become not only a part
of numerous urban transportation systems, but also an effective means of reducing car ownership and the total mileage
(Stillwater et al, 2008). Most initial studies on car sharing in North America dealt with the mobility and flexibility of
the car-sharing system, its influence on the urban transportation system, and ways to expand car-sharing membership;
however, the behavior patterns of car-sharing users was rarely explored in these studies. However, researchers soon
discovered that the characteristics of user behavior were significant to topics studied in the previous works because
behavior patterns may explain some system operation problems from the root itself. Several researchers, for example,
Nobis (2006), Millard-Ball (2006), Celsor and Millard-Ball (2007), Morency (2008, 2009), and Khandlker M. et al
(2011) analyzed usersbehavior patterns, and they all considered it as an important factor determining the success of
the car-sharing system operation.
Thus far, several studies have been conducted outside China on the behavior patterns of car-sharing users. Seik
(2000) analyzed actual car-sharing conditions, investigated the characteristics of travel distance and travel destination
in Singapore, and discovered significant differences between members and non-members. Some studies (2009, 2010)
showed that after joining a car-sharing club, members mainly used car-sharing systems for long holidays or for
shopping and travel; for regular commuting, the members always used public transportation. In the terms of the
frequency and distance of usage, some studies in England showed that 75% of the members used car-sharing less than
5 times per year, that the distance travelled thus by 64% of the members was less than 40 miles. Some studies in
Belgium indicated that used car-sharing less than thrice a month and mainly used it for shopping, visiting friends, and
taking holiday in their leisure time. Millard-Ball (2005) also found that only a small proportion of the members used
car-sharing for regular commuting, and most members used car-sharing for carrying heavy things or for trips with
many destinations. Khandlker et al. (2011) analyzed the empirical data and found that men and French-speaking
members tend to practice car sharing only for short trips, but their usage frequency is high. Meanwhile, it could not be
proved that increasing the number of cars has any effect on maintaining the membership, but it may lead to an increase
in the frequency of usage to some extent.
Unlike in North America, in China, car sharing is only in the initial stage. Because of the lack of practical support,
relatively less research has been carried out on car sharing in China. Huang Zhaoyi (2000) summed up the car-sharing
concept and proposed a number of car-sharing initiatives. Many researchers summarized and analyzed the practice
abroad. Hui Ying (2008) and Ye Liang (2012) opined that by actively encouraging car sharing, households can be
guided into using this alternative rather than buying a car, thus slowing down the trend of households buying a second
car. After reviewing car-sharing practices in cities abroad, Xu Qing (2014) reported various problems related to car
sharing in China. Qie Lisha studied the empty car distribution and created a model to validate this distribution for car-
sharing networks. Although some studies on the behaviors of car-sharing users are noteworthy, most still dealt only
with the level of willingness. Hui Ying (2010, 2012, 2013) and Wang Zengquan (2013) used discrete choice models
for the willingness of joining a car-sharing club and analyzed the potential requirements and possibility of changing
travel behavior.
In China, car-sharing studies just stayed in the level of theoretical analysis and willingness investigation, extremely
lack of the study on members' behavior patterns, which was based on the empirical data. Based on the no-personal
privacy information car rental data of a car-sharing system initiated in July 2013 in Hangzhou, China, this work
analyzed the behavior patterns of long-term users. These users were classified into several groups, and the typical
4668 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 3
usage patterns of behavior in each group were identified and summarized. So far we just have the data about operation
orders, without any data on user profile (e.g. age, income, gender,...and destination of each trip. So this paper only
discussing the behavior patterns of long-term users, which is based on the time of using and distance of one time use.
The results of this work will act as the basis for studying target users and changes in travel behavior, and thus, have a
profound implication on the development and implementation of car sharing in the future.
2.Data source
In July 2013, a car-sharing system started operating in Hangzhou, which was the first massive pilot city for the
project in China. The system was operated by Chelizi Intelligent Technology Company, which had been established
in 2010. Unlike the traditional car rental companies, the node of a car-sharing network is not a store with many workers
but several parking spaces; users can take a car from these spaces and park the car there all by themselves without any
staff supervision. Up to September 2014, 73 nodes of the car-sharing network were in operation. The number of nodes
has continued to increase. At the end of August 2015, the network had 79 nodes and more than 2000s members.
Based on the no-personal privacy information car rental data of the car-sharing operation system, from September
2013 to September 2014, the detailed behavior patterns of long-term users who had registered through the car-sharing
system more than three month ago and had used the system more than once each month were analyzed. According to
the frequency of usage, these uses were classified into several groups. Meanwhile, the typical usage patterns of
behavior in each group were identified, summarized, and compared with those of the all users group. The total number
of orders was 19120. Invalid orders such as revoked orders, orders with unreasonable time points between taking-out
and placing-in the car, and orders with wrong telephone number were eliminated. The number of orders after
eliminating the invalid ones was 14772.
4 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 1. car-sharing network nodes in Hangzhou, China
3. Grouping of users
3.1. Target users determinations
This study analyzed only the effective duration of the members. For example, if the first order of a member was
placed in October 2013 and the last order in January 2014, this member's effective duration was four months.
The total number of users was 1938, and this included two types of usersthe long-term users and occasional users.
The usage period and frequency of usage of occasional users were relatively random. Some of them used the system
for only one or two months. Hence, it was difficult to sum up their regularity. Meanwhile, long-term users had more
stable characteristics of behavior patterns. This group of users contributed just 21% of the total members, but placed
66% of all valid orders. Therefore, long-term users who had used the system for more than three months and whose
frequency of usage was more than once a month (406 members) were the target of this study.
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4669
Author name / Transportation Research Procedia 00 (2017) 000000 3
usage patterns of behavior in each group were identified and summarized. So far we just have the data about operation
orders, without any data on user profile (e.g. age, income, gender,...and destination of each trip. So this paper only
discussing the behavior patterns of long-term users, which is based on the time of using and distance of one time use.
The results of this work will act as the basis for studying target users and changes in travel behavior, and thus, have a
profound implication on the development and implementation of car sharing in the future.
2.Data source
In July 2013, a car-sharing system started operating in Hangzhou, which was the first massive pilot city for the
project in China. The system was operated by Chelizi Intelligent Technology Company, which had been established
in 2010. Unlike the traditional car rental companies, the node of a car-sharing network is not a store with many workers
but several parking spaces; users can take a car from these spaces and park the car there all by themselves without any
staff supervision. Up to September 2014, 73 nodes of the car-sharing network were in operation. The number of nodes
has continued to increase. At the end of August 2015, the network had 79 nodes and more than 2000s members.
Based on the no-personal privacy information car rental data of the car-sharing operation system, from September
2013 to September 2014, the detailed behavior patterns of long-term users who had registered through the car-sharing
system more than three month ago and had used the system more than once each month were analyzed. According to
the frequency of usage, these uses were classified into several groups. Meanwhile, the typical usage patterns of
behavior in each group were identified, summarized, and compared with those of the all users group. The total number
of orders was 19120. Invalid orders such as revoked orders, orders with unreasonable time points between taking-out
and placing-in the car, and orders with wrong telephone number were eliminated. The number of orders after
eliminating the invalid ones was 14772.
4 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 1. car-sharing network nodes in Hangzhou, China
3. Grouping of users
3.1. Target users determinations
This study analyzed only the effective duration of the members. For example, if the first order of a member was
placed in October 2013 and the last order in January 2014, this member's effective duration was four months.
The total number of users was 1938, and this included two types of usersthe long-term users and occasional users.
The usage period and frequency of usage of occasional users were relatively random. Some of them used the system
for only one or two months. Hence, it was difficult to sum up their regularity. Meanwhile, long-term users had more
stable characteristics of behavior patterns. This group of users contributed just 21% of the total members, but placed
66% of all valid orders. Therefore, long-term users who had used the system for more than three months and whose
frequency of usage was more than once a month (406 members) were the target of this study.
4670 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 5
Fig. 2. (a) proportion of target users; (b) proportion of target users' orders
3.2 Classifying target users
Since the frequency of usage varied greatly even among the long-term users, in order to describe the behavior
patterns of different users' precisely, they were classified depending on the interval time between orders, which is the
best index of the frequency of usage. Meanwhile, the number of long-term users fluctuated greatly in terms of monthly
orders and the rapid increase in orders in a peak month would have an adverse impact on the description of the
frequency of usage.
First, the coefficient of fluctuation in peak month orders (the ratio of number of peak month orders to that of off-
peak month) for 406 long-term users was calculated and a scatter diagram, as shown in Fig 3, was plotted. The
coefficient of 2.5 was found to be the boundary in the diagram. The area with coefficient < 2.5 was denser, while that
with coefficient > 2.5 was sparser. Hence, in study, the users whose fluctuant coefficient was more than 2.5 were
considered as the fluctuant users group; the fluctuant coefficient of this group and of stable users group are as listed in
Table 1.
Table 1. Fluctuant coefficients of the peak month
stable users group fluctuant users group
amplitude 1.92 3.17
6 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 3. scatter diagram of the coefficient of fluctuation of peak month orders
Then, the rest of the long-term users were classified according to the frequencies of use. The upper quartile, median,
and lower quartile of order interval time were calculated (Fig 4). In Fig 4, the ordinate is the order interval days. The
green dot and the starting and end points of the vertical lines indicate the median, upper quartile, and lower quartile,
respectively. The lines corresponding to intervals of 2 days and 16 days form the boundaries. In the graph, the region
corresponding to intervals of less than 2 days and more than 16 days are sparse, while the region between intervals of
2 days and 16 days is dense. This region can be further divided at the interval of 8 days. Thus, the stable users were
classified into four groups: the highest frequency users (order intervals mostly within 2 days), higher frequency users
(order interval mostly between 2 days and a week), lower frequency users (order interval mostly between a week and
2 weeks), and lowest frequency users (order interval mostly more than 2 weeks).
The grouping of stable users is described in Table 2. The fluctuant users group and lower frequency users group
had more members, accounting for 37% and 30% of the total members. In the next place, were the members of higher
frequency users group, accounting for 16%, while the numbers of members of the highest frequency users group and
the lowest frequency users group were less.
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4671
Author name / Transportation Research Procedia 00 (2017) 000000 5
Fig. 2. (a) proportion of target users; (b) proportion of target users' orders
3.2 Classifying target users
Since the frequency of usage varied greatly even among the long-term users, in order to describe the behavior
patterns of different users' precisely, they were classified depending on the interval time between orders, which is the
best index of the frequency of usage. Meanwhile, the number of long-term users fluctuated greatly in terms of monthly
orders and the rapid increase in orders in a peak month would have an adverse impact on the description of the
frequency of usage.
First, the coefficient of fluctuation in peak month orders (the ratio of number of peak month orders to that of off-
peak month) for 406 long-term users was calculated and a scatter diagram, as shown in Fig 3, was plotted. The
coefficient of 2.5 was found to be the boundary in the diagram. The area with coefficient < 2.5 was denser, while that
with coefficient > 2.5 was sparser. Hence, in study, the users whose fluctuant coefficient was more than 2.5 were
considered as the fluctuant users group; the fluctuant coefficient of this group and of stable users group are as listed in
Table 1.
Table 1. Fluctuant coefficients of the peak month
stable users group
fluctuant users group
amplitude
1.92
3.17
6 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 3. scatter diagram of the coefficient of fluctuation of peak month orders
Then, the rest of the long-term users were classified according to the frequencies of use. The upper quartile, median,
and lower quartile of order interval time were calculated (Fig 4). In Fig 4, the ordinate is the order interval days. The
green dot and the starting and end points of the vertical lines indicate the median, upper quartile, and lower quartile,
respectively. The lines corresponding to intervals of 2 days and 16 days form the boundaries. In the graph, the region
corresponding to intervals of less than 2 days and more than 16 days are sparse, while the region between intervals of
2 days and 16 days is dense. This region can be further divided at the interval of 8 days. Thus, the stable users were
classified into four groups: the highest frequency users (order intervals mostly within 2 days), higher frequency users
(order interval mostly between 2 days and a week), lower frequency users (order interval mostly between a week and
2 weeks), and lowest frequency users (order interval mostly more than 2 weeks).
The grouping of stable users is described in Table 2. The fluctuant users group and lower frequency users group
had more members, accounting for 37% and 30% of the total members. In the next place, were the members of higher
frequency users group, accounting for 16%, while the numbers of members of the highest frequency users group and
the lowest frequency users group were less.
4672 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 7
Fig. 4. scatter diagram of order interval
Table 2. Grouping of users
Users Group
Characteristic of each group
members
orders
all users
includes all users
1938
14772
long-term users
users who had the used car-sharing
system for more than three months and
whose frequency of usage was more
than once per month
406
(account for 21%
in all users group)
9722
(account for 66%
in all orders)
long-
term
users
group
fluctuant users group
users whose fluctuant coefficient was
more than 2.5
152/37%
3283/33.8%
stable
users
group
highest frequency
users group
the order interval mostly within 2 days
34/8%
2707/27.8%
higher frequency
users group
the order interval mostly between 2
days and a week
63/16%
1814/18.7%
lower frequency
users group
the order interval mostly between 1
week and 2 weeks
120/30%
1646/16.9%
lowest frequency
users group
the order interval mostly more than 2
weeks
37/9%
271/2.8%
Label: In the fluctuant, highest frequency users, higher frequency of users, lower frequency users, and lowest frequency users groups, the ratio of
members and orders are shown with respect to the long-term users group.
4 Analysis of the behavior patterns of target users
4.1 New members
Members of all users group maintained a relatively stable growth trend, despite a decrease in February 2014 and a
rebound thereafter. The stable users had 35 new users; this number is less than the number of new users in the all users
group (about 120 users per month). Since March 2014, the number of new members in the stable users group gradually
diminished. Further analysis of the new members of this group showed that the highest frequency users and lowest
frequency users groups both had less than 5 new users per month, while the lower frequency users and fluctuant users
groups had much more new members each month; however, the trends of the five groups of stable users were basically
the same.
8 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 5. (a) new members of all users group; (b) new members of long-term users group
Label: In this study, the month of users’ first order is defined as the month of new member. Since the study only included data from September
2013 to September 2014, if a user’s first order was before September 2013, this paper still considered month of the user becoming a new member
as September 2013. Because of the method of analysis, the data for September 2013 yielded a cumulative result in the early months, and after July
2014, there were no new members in the stable users group. Hence, this figure only includes the new member data from October 2013 to June 2014.
4.2 Monthly orders
Monthly orders of all users group maintained a growth ratio of 4.5%, despite of a decrease in growth in October
2013 and February 2014. The increase in orders of the long-term users group was slower than that of the all users
group. Before July 2014, the trend of the number of orders of the long-term users was the same as that of the all users
group. However, after July 2014, the number of orders of the long-term users declined. This transformation may be
related to the change in the number of new members.
The stability of the monthly orders of long-term users was in sharp contrast with the fluctuations of the monthly
orders of all users. The obvious decline of the all users group in October 2013 and February 2014 was much sharper
than that of the long-term users group. Meanwhile, in the target year of study, there were two long legal holidays,
namely, the National Day in October 2013 and the Spring Festival in February 2014. This sharp contrast may be a
result of the different sensitivities attributed to the holiday. The members of the all users group were more sensitive to
the long-term holiday than were the long-term users, for the former group included some occasional users.
The monthly orders of five stable users group maintained a stable slow growth despite the changes in February and
March 2014. There was a little decline, which may be attributed to the decrease in the number of new members in
February 2014. Although that month coincided with the Spring Festival, it is considered that this little decline was not
caused by the legal holiday because the National Day did not affect the orders in October 2013.
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4673
Author name / Transportation Research Procedia 00 (2017) 000000 7
Fig. 4. scatter diagram of order interval
Table 2. Grouping of users
Users Group
Characteristic of each group
members
orders
all users
includes all users
1938
14772
long-term users
users who had the used car-sharing
system for more than three months and
whose frequency of usage was more
than once per month
406
(account for 21%
in all users group)
9722
(account for 66%
in all orders)
long-
term
users
group
fluctuant users group
users whose fluctuant coefficient was
more than 2.5
152/37%
3283/33.8%
stable
users
group
highest frequency
users group
the order interval mostly within 2 days
34/8%
2707/27.8%
higher frequency
users group
the order interval mostly between 2
days and a week
63/16%
1814/18.7%
lower frequency
users group
the order interval mostly between 1
week and 2 weeks
120/30%
1646/16.9%
lowest frequency
users group
the order interval mostly more than 2
weeks
37/9%
271/2.8%
Label: In the fluctuant, highest frequency users, higher frequency of users, lower frequency users, and lowest frequency users groups, the ratio of
members and orders are shown with respect to the long-term users group.
4 Analysis of the behavior patterns of target users
4.1 New members
Members of all users group maintained a relatively stable growth trend, despite a decrease in February 2014 and a
rebound thereafter. The stable users had 35 new users; this number is less than the number of new users in the all users
group (about 120 users per month). Since March 2014, the number of new members in the stable users group gradually
diminished. Further analysis of the new members of this group showed that the highest frequency users and lowest
frequency users groups both had less than 5 new users per month, while the lower frequency users and fluctuant users
groups had much more new members each month; however, the trends of the five groups of stable users were basically
the same.
8 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 5. (a) new members of all users group; (b) new members of long-term users group
Label: In this study, the month of users’ first order is defined as the month of new member. Since the study only included data from September
2013 to September 2014, if a user’s first order was before September 2013, this paper still considered month of the user becoming a new member
as September 2013. Because of the method of analysis, the data for September 2013 yielded a cumulative result in the early months, and after July
2014, there were no new members in the stable users group. Hence, this figure only includes the new member data from October 2013 to June 2014.
4.2 Monthly orders
Monthly orders of all users group maintained a growth ratio of 4.5%, despite of a decrease in growth in October
2013 and February 2014. The increase in orders of the long-term users group was slower than that of the all users
group. Before July 2014, the trend of the number of orders of the long-term users was the same as that of the all users
group. However, after July 2014, the number of orders of the long-term users declined. This transformation may be
related to the change in the number of new members.
The stability of the monthly orders of long-term users was in sharp contrast with the fluctuations of the monthly
orders of all users. The obvious decline of the all users group in October 2013 and February 2014 was much sharper
than that of the long-term users group. Meanwhile, in the target year of study, there were two long legal holidays,
namely, the National Day in October 2013 and the Spring Festival in February 2014. This sharp contrast may be a
result of the different sensitivities attributed to the holiday. The members of the all users group were more sensitive to
the long-term holiday than were the long-term users, for the former group included some occasional users.
The monthly orders of five stable users group maintained a stable slow growth despite the changes in February and
March 2014. There was a little decline, which may be attributed to the decrease in the number of new members in
February 2014. Although that month coincided with the Spring Festival, it is considered that this little decline was not
caused by the legal holiday because the National Day did not affect the orders in October 2013.
4674 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 9
Fig. 6. monthly orders of each group
4.3 Weekdays and weekends
The orders were classified according to weekdays and weekends. It was clear that the data of the all users, long-
term users, and fluctuant users groups showed little difference, and the daily ratio for each group was close to the
average of four stable groups for that day.
The times of usage of these four stable users group were different. The lower frequency of the users group was
more likely to use car-sharing over the weekend, and the ratio members of using car-sharing on Saturday was much
more than that for Sunday. Meanwhile, the higher frequency of the users group preferred to use car sharing on
weekdays. Orders on weekdays were analyzed in detail, and the variance of each group was calculated. Among the
four stable users group, the higher the frequency of users, the smaller was the variance of the group. That is, the higher
the frequency of users, the more likely it was the number of daily orders during weekdays was the same. Thus, it was
found that there was a small trend of traffic commuting in the highest frequency users group.
Table 3. Variance of the ratio of orders in weekdays
Group
Highest
frequency
stability users
Higher frequency
stability users
Lower frequency
stability users
Lowest frequency
stability users
Fluctuant
users All users Long-term
users
Variance 0.000144 0.000341 0.00062 0.001053 0.000282 0.000153 0.000228
10 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 7. daily ratios of each group
4.4 Single-use time
The relative graph between the single-use time and ratio of orders was plotted as shown in Fig 8, considering units
of day and hour. The curves of all groups almost coincide. More than 90% of the users used car-sharing within one
day, and the single-use time of only a few (less than 2%) users was more than 3 days.
The orders for which single-use time was within 1 day was analyzed on an hourly basis. Even in this case, the
curves for all the groups were found to coincide. The ratio of single-use time peaked at 13 hours, and peak lasted for
0–6 hours. Meanwhile, the ratio of single-use time reached a secondary peak, which was approximately a quarter of
the main peak, at 1516 hours, and its peak lasted for 1218 hours. The only difference between these four stable users
group was in terms of the height of the main peak. The higher the frequency of the users, the higher was the main
peak. The height of the main peak of the highest frequency users group was 21.6%, while that of the lowest frequency
users group was 13.94%.
Car-sharing thus forms complements the taxi and traditional car rental services, and 95% of all taxi orders have a
single-use time of 30 min. Meanwhile, for the traditional car rental service, a day is considered as a unit. These
diverse services enrich the overall traffic modes of the transportation system.
Fig. 8. (a) single-use time of each group with unit of day; (b) single-use time of each group with unit of hour
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4675
Group
Highest
frequency
stability users
Higher frequency
stability users
Lower frequency
stability users
Lowest frequency
stability users
Fluctuant
users
All users
Long-term
users
Variance
0.000144
0.000341
0.00062
0.001053
0.000282
0.000153
0.000228
10 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 7. daily ratios of each group
4.4 Single-use time
The relative graph between the single-use time and ratio of orders was plotted as shown in Fig 8, considering units
of day and hour. The curves of all groups almost coincide. More than 90% of the users used car-sharing within one
day, and the single-use time of only a few (less than 2%) users was more than 3 days.
The orders for which single-use time was within 1 day was analyzed on an hourly basis. Even in this case, the
curves for all the groups were found to coincide. The ratio of single-use time peaked at 13 hours, and peak lasted for
0–6 hours. Meanwhile, the ratio of single-use time reached a secondary peak, which was approximately a quarter of
the main peak, at 1516 hours, and its peak lasted for 12–18 hours. The only difference between these four stable users
group was in terms of the height of the main peak. The higher the frequency of the users, the higher was the main
peak. The height of the main peak of the highest frequency users group was 21.6%, while that of the lowest frequency
users group was 13.94%.
Car-sharing thus forms complements the taxi and traditional car rental services, and 95% of all taxi orders have a
single-use time of 30 min. Meanwhile, for the traditional car rental service, a day is considered as a unit. These
diverse services enrich the overall traffic modes of the transportation system.
Fig. 8. (a) single-use time of each group with unit of day; (b) single-use time of each group with unit of hour
4676 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 11
4.5 Taking-out and placing-in of cars
The taking-out time of the all users group and long-term users group were basically the same; similarly, the placing-
in time for both groups were same too. The taking-out time reached a peak at 7:00 hours and began to decline 19:00
hours. Meanwhile, the placing-in time had two peaks—6:00 hours and the 16:00 hours.
The taking-out time and placing-in time for the four stable users groups all started at 6:00 hours, but their ending
times were different. The taking-out and placing-in times of the highest frequency users group had best concentricity
and was more in line with the time of commuting traffic (7:00 hours to 20:00 hours). Meanwhile, the lowest frequency
users group had better dispersity. In addition to the time before shut down (1:00 to 5:00 hours), the ratio of taking out
a car and placing in a car was high. The higher frequency users and lower frequency users groups had taking-out and
placing-times intermediate those of the highest frequency users group and lowest frequency users group. The
characteristic of the fluctuant users group was similar to that of the lower frequency users group.
Fig. 9. (a) taking-out and placing-in time of all users; (b) taking-out and placing-in time of long-term users
Fig. 9. (c) taking-out and placing-in time of highest frequency users; (b) taking-out and placing-in time of higher frequency users
12 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 9. (e) taking-out and placing-in time of lower frequency users; (f) taking-out and placing-in time of lowest frequency users
Fig. 9. (g) taking-out and placing-in time of fluctuant users
4.6 Single-use distance
The scale graphs of single-use distance and orders’ proportion of all groups also essentially coincided. Peaks of the
graphs were between 20 and 30 km, and its fastigium was located in the range 1–70 km, which contained more than
80% orders. The only difference was the obvious increase in the ratio of the highest frequency users group for the
short distance (120 km).
In order to describe the characteristics of the highest frequency users better, the cost of taxi was compared with that
of car-sharing for short distance travel (Fig 10) by considering 8 km for the price equilibrium point. In another words,
within 8 km, selecting a taxi was cheaper than car sharing. Under this condition, the members of the highest frequency
users group were still willing to choose car sharing. Thus, combined with their single-use time, it is obvious that the
members of the highest frequency users group had the trend of short time and short distance travel, which was quite
similar to commuting travel, and this trend may develop into a habit that cannot be changed easily in the future.
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4677
Author name / Transportation Research Procedia 00 (2017) 000000 11
4.5 Taking-out and placing-in of cars
The taking-out time of the all users group and long-term users group were basically the same; similarly, the placing-
in time for both groups were same too. The taking-out time reached a peak at 7:00 hours and began to decline 19:00
hours. Meanwhile, the placing-in time had two peaks—6:00 hours and the 16:00 hours.
The taking-out time and placing-in time for the four stable users groups all started at 6:00 hours, but their ending
times were different. The taking-out and placing-in times of the highest frequency users group had best concentricity
and was more in line with the time of commuting traffic (7:00 hours to 20:00 hours). Meanwhile, the lowest frequency
users group had better dispersity. In addition to the time before shut down (1:00 to 5:00 hours), the ratio of taking out
a car and placing in a car was high. The higher frequency users and lower frequency users groups had taking-out and
placing-times intermediate those of the highest frequency users group and lowest frequency users group. The
characteristic of the fluctuant users group was similar to that of the lower frequency users group.
Fig. 9. (a) taking-out and placing-in time of all users; (b) taking-out and placing-in time of long-term users
Fig. 9. (c) taking-out and placing-in time of highest frequency users; (b) taking-out and placing-in time of higher frequency users
12 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
Fig. 9. (e) taking-out and placing-in time of lower frequency users; (f) taking-out and placing-in time of lowest frequency users
Fig. 9. (g) taking-out and placing-in time of fluctuant users
4.6 Single-use distance
The scale graphs of single-use distance and orders’ proportion of all groups also essentially coincided. Peaks of the
graphs were between 20 and 30 km, and its fastigium was located in the range 1–70 km, which contained more than
80% orders. The only difference was the obvious increase in the ratio of the highest frequency users group for the
short distance (1–20 km).
In order to describe the characteristics of the highest frequency users better, the cost of taxi was compared with that
of car-sharing for short distance travel (Fig 10) by considering 8 km for the price equilibrium point. In another words,
within 8 km, selecting a taxi was cheaper than car sharing. Under this condition, the members of the highest frequency
users group were still willing to choose car sharing. Thus, combined with their single-use time, it is obvious that the
members of the highest frequency users group had the trend of short time and short distance travel, which was quite
similar to commuting travel, and this trend may develop into a habit that cannot be changed easily in the future.
4678 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 13
Fig. 10. single-use distance of each group
Fig. 11. price of taxi and car-sharing for short distance travel
Label: 1. The price of taxi referred was obtained by referring to taxi operators’ unified pricing in Hangzhou.
2. The price of car-sharing was obtained by referring to the official website and chosen vehicle type (Buick), which was chosen more than
60% of the members.
4.7 Fluctuant Users
Fluctuant users were studied intensively based on the curve of monthly orders. It was found that only peak month
existed, and there was no low ebb month. The orders’ indexes in the peak month and off-peak month were compared
14 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
among the four stable users group (Table 3) in terms of monthly orders, frequency of usage, orders’ ratio for weekdays
and weekends, single-use time, and the time of taking-out and placing-in the car.
Table 4. Characteristics of peak month orders and off-peak month orders
Peak month orders
Off-peak month orders
Frequency of Usage
7.53 (similar to higher frequency users group)
2.02 (similar to lower frequency users group)
Orders’ ratio of weekdays
and weekends
the daily ratio was same within each group, but
differed among the four stable users group
similar to higher frequency users group
Single-use time
similar to lower frequency users group
similar to lower frequency users group
Time of taking-out and
placing-in the car
similar to higher frequency users group
similar to lower frequency users group
Members of the fluctuant users group had frequency of usage similar to that of lower frequency users, but with
higher fluctuation. However, this study did not consider behaviors such as the factor of this fluctuation, whether this
fluctuation can be influenced, and whether this fluctuation will become stable or fluctuate forever in the future; these
matters need to be discussed in the future.
4.8 Summary
The characteristics of all users, long-term users, fluctuant users, and four stable users groups were analyzed mainly
in terms of new members, monthly orders, orders’ ratio of weekdays and weekend, sing-use time, time of taking-out
and placing-in car, and single-use distance. Through comparisons, it was found that the members of all users group
maintained a growing trend, which was faster than that of the long-term users group. The numbers of orders of the all
users group was more sensitive than that of the long-term users group.
Members of the four stable users group were analyzed in detail. The difference among them was obvious. The
higher the frequency of usage, the more likely the group was to use car sharing during weekdays. In contrast, the users
with lower frequency of usage preferred to use car-sharing services over the weekend. With regard the time of taking-
out and placing-in the car, the higher the frequency of usage, the better was the centrality of single-use time. The times
for lower frequency users were more discrete. Meanwhile, the curves of single-use time and single-use distance were
basically the same. The only difference was that the higher the frequency of the users, the higher was the proportion
of the orders of short time travel (1 to 3 h). In addition, the ratio of short distance travel for the highest frequency users
group was obviously larger than that for the other three groups. The different characteristics of the four stable users
group were plotted in a radar graph (Fig 12), and the numbers 1 to 4 were used to represent the significance level from
weak to strong. The typical users’ behavior patterns were summarized as follows:
Highest Frequency Users Group: Members' monthly orders were maintained at a high but stable level, and the
interval time of great majority of orders was within 2 days. From the viewpoint of single-use period, compared with
members of the other three stable groups, there was a higher proportion of using car-sharing services on weekdays,
and the ratio of daily orders was almost the same from Monday to Friday. Their single-use time was concentrated from
7:00 to 20:00 hours, and the proportion of orders at other times was very low. Meanwhile, the short time and short
distance travel was more likely unlike in the case of the other three stable groups. Overall, the travel behavior of the
members of this group was closely related to commuting traffic.
Lowest Frequency Users Group: Members' monthly orders were maintained at a low but stable level, and the
interval time for a great majority of the orders was between 2 weeks and a month. With regard single-use period,
compared with the members of other three stable groups, there was a higher proportion of using car-sharing services
over the weekend, and the ratio of orders on Saturday was larger than that on Sunday. Their single-use time was
disperse. In addition to the time from 1:00 to 17:00 hours, the proportion of orders at other times were maintained at
a higher level, and there was no obvious fluctuation. Meanwhile, the orders for short time and short distance travel
were the least among the four stable groups. Overall, the travel behavior of these users was closely related to living
trips, and they showed little tendency of commuting traffic.
The characteristics of the higher frequency users and lower frequency users groups were intermediate those of the
highest frequency users and lowest frequency users groups. The behavior of the higher frequency users group was
more similar to the highest frequency users group, while that of the lower frequency users group was more similar to
that of the lowest frequency users group.
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4679
Author name / Transportation Research Procedia 00 (2017) 000000 13
Fig. 10. single-use distance of each group
Fig. 11. price of taxi and car-sharing for short distance travel
Label: 1. The price of taxi referred was obtained by referring to taxi operators’ unified pricing in Hangzhou.
2. The price of car-sharing was obtained by referring to the official website and chosen vehicle type (Buick), which was chosen more than
60% of the members.
4.7 Fluctuant Users
Fluctuant users were studied intensively based on the curve of monthly orders. It was found that only peak month
existed, and there was no low ebb month. The orders’ indexes in the peak month and off-peak month were compared
14 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
among the four stable users group (Table 3) in terms of monthly orders, frequency of usage, orders’ ratio for weekdays
and weekends, single-use time, and the time of taking-out and placing-in the car.
Table 4. Characteristics of peak month orders and off-peak month orders
Peak month orders
Off-peak month orders
Frequency of Usage
7.53 (similar to higher frequency users group)
2.02 (similar to lower frequency users group)
Orders’ ratio of weekdays
and weekends
the daily ratio was same within each group, but
differed among the four stable users group
similar to higher frequency users group
Single-use time
similar to lower frequency users group
similar to lower frequency users group
Time of taking-out and
placing-in the car
similar to higher frequency users group
similar to lower frequency users group
Members of the fluctuant users group had frequency of usage similar to that of lower frequency users, but with
higher fluctuation. However, this study did not consider behaviors such as the factor of this fluctuation, whether this
fluctuation can be influenced, and whether this fluctuation will become stable or fluctuate forever in the future; these
matters need to be discussed in the future.
4.8 Summary
The characteristics of all users, long-term users, fluctuant users, and four stable users groups were analyzed mainly
in terms of new members, monthly orders, orders’ ratio of weekdays and weekend, sing-use time, time of taking-out
and placing-in car, and single-use distance. Through comparisons, it was found that the members of all users group
maintained a growing trend, which was faster than that of the long-term users group. The numbers of orders of the all
users group was more sensitive than that of the long-term users group.
Members of the four stable users group were analyzed in detail. The difference among them was obvious. The
higher the frequency of usage, the more likely the group was to use car sharing during weekdays. In contrast, the users
with lower frequency of usage preferred to use car-sharing services over the weekend. With regard the time of taking-
out and placing-in the car, the higher the frequency of usage, the better was the centrality of single-use time. The times
for lower frequency users were more discrete. Meanwhile, the curves of single-use time and single-use distance were
basically the same. The only difference was that the higher the frequency of the users, the higher was the proportion
of the orders of short time travel (1 to 3 h). In addition, the ratio of short distance travel for the highest frequency users
group was obviously larger than that for the other three groups. The different characteristics of the four stable users
group were plotted in a radar graph (Fig 12), and the numbers 1 to 4 were used to represent the significance level from
weak to strong. The typical users’ behavior patterns were summarized as follows:
Highest Frequency Users Group: Members' monthly orders were maintained at a high but stable level, and the
interval time of great majority of orders was within 2 days. From the viewpoint of single-use period, compared with
members of the other three stable groups, there was a higher proportion of using car-sharing services on weekdays,
and the ratio of daily orders was almost the same from Monday to Friday. Their single-use time was concentrated from
7:00 to 20:00 hours, and the proportion of orders at other times was very low. Meanwhile, the short time and short
distance travel was more likely unlike in the case of the other three stable groups. Overall, the travel behavior of the
members of this group was closely related to commuting traffic.
Lowest Frequency Users Group: Members' monthly orders were maintained at a low but stable level, and the
interval time for a great majority of the orders was between 2 weeks and a month. With regard single-use period,
compared with the members of other three stable groups, there was a higher proportion of using car-sharing services
over the weekend, and the ratio of orders on Saturday was larger than that on Sunday. Their single-use time was
disperse. In addition to the time from 1:00 to 17:00 hours, the proportion of orders at other times were maintained at
a higher level, and there was no obvious fluctuation. Meanwhile, the orders for short time and short distance travel
were the least among the four stable groups. Overall, the travel behavior of these users was closely related to living
trips, and they showed little tendency of commuting traffic.
The characteristics of the higher frequency users and lower frequency users groups were intermediate those of the
highest frequency users and lowest frequency users groups. The behavior of the higher frequency users group was
more similar to the highest frequency users group, while that of the lower frequency users group was more similar to
that of the lowest frequency users group.
4680 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 15
Members of the fluctuant users group showed frequent usage in off-peak months at rates of twice per month, and
this is similar to the behavior of lower frequency users, but with a large fluctuation in the peak month. Their
characteristics were somewhat similar to those of the members of the higher frequency users group and lower
frequency users group, though the similarity was not very obvious.
Fig. 12. different characteristics of the four stable users groups
5 conclusions
Car sharing, which is an innovative transportation method, is in its initial stage in China. The studies on car sharing
until date were mostly made at a theoretical level and did not consider the characteristics of user behavior patterns.
After a detailed analysis of one-year data of car-sharing (the data was a kind of car rental data without personal privacy
information) in Hangzhou, it was found that orders of the long-term users group were more stable and less sensitive
to legal holidays than those of the other groups. Meanwhile, the number of long-term users tended to be stable, while
the number of new members of the all users group still maintained an increasing ratio of 4.5%. Other characteristics,
such as single-use time and single-use distance showed, were slightly different among the all users and long-term users
groups.
Long-term users were studied intensively based on the difference of user behavior pattern in terms of four stable
users group. The behavior patterns of the highest frequency users group (for example: order interval was almost within
2 days; a higher proportion of using car-sharing on weekdays; single-use time was concentrated from 7:00 to 20:00
hours; the short time and short distance travel was more likely) were similar to the characteristics of commuting travel.
To a great extent, these members considered car-sharing only as a substitute for private car or public transportation,
and maybe they use car-sharing only as a transition phase before buying their own car. Abroad, car-sharing was
established when families had more than one private car, and members always used car-sharing to go shopping or for
entertainment in their leisure time. Owing to the different conditions between China and other countries, few users
abroad use car sharing in a manner similar to the members of the highest frequency users in China. Thus, considering
the need of families to purchase private cars, several issues still need to be studied intensively. For example: Whether
this members lead to an increase in the usage of cars and cause an increase in the burden on the transportation system;
and whether car sharing promotes potential users to buy their private car. Meanwhile, from the viewpoint of a typical
transportation system, further study on whether such users needed to be given some positive policies to maintain their
car-sharing habit is also necessary.
16 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
On a separate note, the lowest frequency users (characteristics, for example: interval time of a great majority of
orders was between 2 weeks and a month; a higher proportion of using car-sharing over the weekend; and single-use
time was disperse) enjoyed the mobility of car-sharing resources and thus decreased private car ownership effectively.
Because of the dispersity of the single-use time, these users do not increase the burden on the transportation system.
In fact, these members are more similar to the car-sharing users abroad. However, there are some differences between
China and abroad in that whether such users are the major users of car-sharing system and whether positive policies
should be in place for them; these matters should be studied in the future.
Meanwhile, members of the fluctuant users group are a type of special users. Their frequency of usage in the off-
peak month was low, about three times less than that in the peak month. The single-use time of these users was disperse,
and the possibility of being influenced was not clear now. Hence, how to lead the behavior patterns of these users
through policy, and how to balance the mobility of travel and reduce the burden on the transportation system still need
to be analyzed through questionnaire surveys.
This work analyzed the behavior patterns of long-term users of car sharing. The study will be a foundation of
questionnaire surveys and political analysis. Car sharing, which is an innovative transportation method, has been
attempted in China. Whether car sharing will reduce the rate of purchase of private cars by families and will change
the users' travel behavior patterns still need to be investigated in detail. Meanwhile, with the development of the car-
sharing system, balancing enterprise efficiency and the profits of citizen transportation system also need to be
considered in the long term.
Acknowledgments
The data of this work was provided by "Chefenxiang" project, which was operated by Chelizi Intelligent Technology
Company. The data was a kind of car rental data, without personal privacy information. Many thanks to the their
workers for their patient introduction about operation system and detailed explanation to our questions about the data
on operation orders.
Reference
Meijkamp, R., 1998. Changing consumer behaviour through eco-efficient services: an empirical study of carsharing in the Netherlands. Business
Strategy and the Environment 7 (4), 234244.
Ohta, H., Fujii, S., Nishimura, Y., 2009. Psychological analysis of acceptance of pro-environmental use of automobile: cases of carsharing and eco-
car. Paper Presented at the 88th Annual Meeting of Transportation Research Board, Washington, DC.
Shaheen, S.A., Cohen, A.P., Chung, M.S., 2009.North American car sharing: a ten-year retrospective. Paper Presented at the 88th Annual Meeting
of Transportation Research Board, Washington, DC.
Stillwater, T., Mokhtarian, P.L., Shaheen, S., 2008. Car sharing and built environment: a GIS-based study of one US operator. Paper Presented at
the 87th Annual Meeting of Transportation Research Board, Washington, DC.
Nobis, C., 2006. Car Sharing as a Key Contribution to Multimodal and Sustainable Mobility Behaviour The Situation of Carsharing in Germany.
Transportation Research Record 1986, pp. 8997.
Millard-Ball, A., Morray, G., Schure, J.T., 2006. Car-sharing as a parking management strategy. Paper Presented at the 85th Annual Meeting of
Transportation Research Board, Washington, DC.
Celsor, C., Millard-Ball, A., 2007. Where Does Carsharing Work? Using GIS to Assess Market Potential. Transportation Research Record 1992,
pp. 6169.
Morency, C., Trepanier, M., Basile, M., 2008. Object-Oriented Analysis of Car sharing System. Transportation Research Record 2063, pp. 105
112.
Morency, C., g, K.M.N., g, V., Islam, M.T., 2009. Application of a dynamic ordered profit model for predicting the activity persistency of car
sharing member. Paper Presented at the 88th Annual Meeting of Transportation Research Board, Washington, DC.
M. Catherine Morency, Mohammed Tazul Islam, Vincent Grasset.,2011. Modeling users behavior of a car sharing program: Application of a joint
hazard and zero inflated dynamic ordered probability model. Transportation Research Part A 46 (2012) 241254
Foo Tuan Seik,2000. Vehicle ownership restraints and car sharing in Singapore. Habitat Intemational,(24):75 90
Customer Survey Car-Sharing Brussels. Project MOMO Grant agreement, 2009.
Report Survey on Satisfaction of Cambio Client in Wallonia. MOMO Grant agreement,2010
Millard Ball A., 2005. Car sharing: Where and how it succeeds. Washington DC: Transportation Research Board.
Muheim, P. and Partner.,1996. Car Sharing Studies: An Investigation, Prepared for Graham Lightfoot, Ireland..
Muheim, P. and Partner.,1996. Car Sharing Studies: An Investigation, Lucerne, Switzerland; Prepared for Balance Services AG & Graham
Lightfoot, EU-SAVE Ireland:156~158
Zhaoyi Huang, Dongyuan Yang, 2000. The development condition of car sharing abroad. Urban Planning Forum, 2000(6): 50-55.
Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682 4681
16 Mengtao DING etc/ Transportation Research Procedia 00 (2017) 000000
On a separate note, the lowest frequency users (characteristics, for example: interval time of a great majority of
orders was between 2 weeks and a month; a higher proportion of using car-sharing over the weekend; and single-use
time was disperse) enjoyed the mobility of car-sharing resources and thus decreased private car ownership effectively.
Because of the dispersity of the single-use time, these users do not increase the burden on the transportation system.
In fact, these members are more similar to the car-sharing users abroad. However, there are some differences between
China and abroad in that whether such users are the major users of car-sharing system and whether positive policies
should be in place for them; these matters should be studied in the future.
Meanwhile, members of the fluctuant users group are a type of special users. Their frequency of usage in the off-
peak month was low, about three times less than that in the peak month. The single-use time of these users was disperse,
and the possibility of being influenced was not clear now. Hence, how to lead the behavior patterns of these users
through policy, and how to balance the mobility of travel and reduce the burden on the transportation system still need
to be analyzed through questionnaire surveys.
This work analyzed the behavior patterns of long-term users of car sharing. The study will be a foundation of
questionnaire surveys and political analysis. Car sharing, which is an innovative transportation method, has been
attempted in China. Whether car sharing will reduce the rate of purchase of private cars by families and will change
the users' travel behavior patterns still need to be investigated in detail. Meanwhile, with the development of the car-
sharing system, balancing enterprise efficiency and the profits of citizen transportation system also need to be
considered in the long term.
Acknowledgments
The data of this work was provided by "Chefenxiang" project, which was operated by Chelizi Intelligent Technology
Company. The data was a kind of car rental data, without personal privacy information. Many thanks to the their
workers for their patient introduction about operation system and detailed explanation to our questions about the data
on operation orders.
Reference
Meijkamp, R., 1998. Changing consumer behaviour through eco-efficient services: an empirical study of carsharing in the Netherlands. Business
Strategy and the Environment 7 (4), 234244.
Ohta, H., Fujii, S., Nishimura, Y., 2009. Psychological analysis of acceptance of pro-environmental use of automobile: cases of carsharing and eco-
car. Paper Presented at the 88th Annual Meeting of Transportation Research Board, Washington, DC.
Shaheen, S.A., Cohen, A.P., Chung, M.S., 2009.North American car sharing: a ten-year retrospective. Paper Presented at the 88th Annual Meeting
of Transportation Research Board, Washington, DC.
Stillwater, T., Mokhtarian, P.L., Shaheen, S., 2008. Car sharing and built environment: a GIS-based study of one US operator. Paper Presented at
the 87th Annual Meeting of Transportation Research Board, Washington, DC.
Nobis, C., 2006. Car Sharing as a Key Contribution to Multimodal and Sustainable Mobility Behaviour The Situation of Carsharing in Germany.
Transportation Research Record 1986, pp. 8997.
Millard-Ball, A., Morray, G., Schure, J.T., 2006. Car-sharing as a parking management strategy. Paper Presented at the 85th Annual Meeting of
Transportation Research Board, Washington, DC.
Celsor, C., Millard-Ball, A., 2007. Where Does Carsharing Work? Using GIS to Assess Market Potential. Transportation Research Record 1992,
pp. 6169.
Morency, C., Trepanier, M., Basile, M., 2008. Object-Oriented Analysis of Car sharing System. Transportation Research Record 2063, pp. 105
112.
Morency, C., g, K.M.N., g, V., Islam, M.T., 2009. Application of a dynamic ordered profit model for predicting the activity persistency of car
sharing member. Paper Presented at the 88th Annual Meeting of Transportation Research Board, Washington, DC.
M. Catherine Morency, Mohammed Tazul Islam, Vincent Grasset.,2011. Modeling users behavior of a car sharing program: Application of a joint
hazard and zero inflated dynamic ordered probability model. Transportation Research Part A 46 (2012) 241254
Foo Tuan Seik,2000. Vehicle ownership restraints and car sharing in Singapore. Habitat Intemational,(24):75 90
Customer Survey Car-Sharing Brussels. Project MOMO Grant agreement, 2009.
Report Survey on Satisfaction of Cambio Client in Wallonia. MOMO Grant agreement,2010
Millard Ball A., 2005. Car sharing: Where and how it succeeds. Washington DC: Transportation Research Board.
Muheim, P. and Partner.,1996. Car Sharing Studies: An Investigation, Prepared for Graham Lightfoot, Ireland..
Muheim, P. and Partner.,1996. Car Sharing Studies: An Investigation, Lucerne, Switzerland; Prepared for Balance Services AG & Graham
Lightfoot, EU-SAVE Ireland:156~158
Zhaoyi Huang, Dongyuan Yang, 2000. The development condition of car sharing abroad. Urban Planning Forum, 2000(6): 50-55.
4682 Ying HUI et al. / Transportation Research Procedia 25C (2017) 4666–4682
Author name / Transportation Research Procedia 00 (2017) 000000 17
Kaixuan Xia, Shengming He, 2006. The theory and practice of car-sharing service abroad. City Issues, 2006(4): 87-92.
Yue Xue, Tongyu Yang, Sunbin Wen, 2008. Social characteristics and develop modes of car sharing. Techno economics & Management Research,
2008(1):54-58
Ying Hui, 2008. Research on traffic development and planning of historic blocks in the context of mobility. Tongji University.
Liang Ye, Dongyuan Yang, 2010. The development and application research of car-sharing and motorized management. China Population Research
and Environment, 2010(20):243-247.
Xu Qing, Yang Dongyuan, Hui Ying, et al.,2014. The car sharing in China is on the way. Traffic and Transportation, 2014(12):121-124.
Lisha Qie, 2007. Study of empty car allocation based on the "car-sharing". Southwest Jiao Tong University.
Ying Hui, Mingquan Wang.,2010. Consuming demand incentive of potential car sharing users and its developing policytake Shanghai as a case
study (EI). 2010 International Conference on Intelligent Computation Technology and Automation, 1031-1034, Changsha.
Miao Zhang, Ying Hui, Mingquan Wang, 2012. Car sharing impact and urban motorization The Seventh China Intelligent Transportation Annual
Meeting Outstanding Papers-Intelligent Transportation Application. pp:673-678
Yanhua Song, 2013. Study on the influence of individual travel mode on the willingness of joining car sharing. Tongji University.
Wang M, Martin E, Shaheen S.,2012. Car sharing in Shanghai, China: analysis of behavior-ral response to a local survey and potential
competition[C]. Transportation Research Record. Washington D.C.,
LI Yan-hong , YUAN Zhen-zhou , XIE Hai-hong et.al.,2007. Analysis on Trips Characteristics of Taxi in Suzhou Based on OD Data. Journal of
Transportation Systems Engineering and Information Technology. 2007.10(7):85-89
... The four types of carsharing travel characteristics are summarised in Table 5. In previous research, long-term carsharing users' fastigium of single-use distance was located in the range 1-70 km, which contained more than 80% of orders [41], which is consistent with the research results of this paper. Meanwhile, scholars have found that the higher frequency carsharing user groups prefer to use carsharing on weekdays. ...
... Meanwhile, scholars have found that the higher frequency carsharing user groups prefer to use carsharing on weekdays. The lower frequency carsharing user groups are more likely to use carsharing over the weekend [41]. Similarly, in this paper, carsharing travellers were also divided into travelling on workdays and weekends (LDLE: weekends; MSDBC, MSDBR and LDBR: workdays). ...
Article
Full-text available
As a new urban travel mode, carsharing is significantly different from private cars, buses and other travel modes. Therefore, clarifying the typical characteristics of carsharing, such as individual users’ attributes, travel environment and travel behaviour, is conducive to accurately grasping the development of carsharing. In this study, a selective ensemble learning model is established to analyse typical travel characteristics of carsharing. Firstly, personal characteristics, environmental characteristics and behavioural characteristics were obtained through integrating order data, global positioning system data and station information. Then, based on a consolidated view of carsharing, different types of carsharing travel characteristics were distinguished using selective ensemble learning. Lastly, all kinds of carsharing travel are described in detail. It was identified through this research that carsharing travel can be divided into four kinds: long distance for leisure and entertainment, medium and short distances for business and commuting, a mixed category of medium and short distances for business and residence, and a mixed category of long distance for business and residence. This study can provide a theoretical reference and practical basis for precise planning and design and the scientific operation of carsharing.
... ere are divergences in the relationship between shared mobility and private car purchase. Some hold that shared mobility can shift individuals' dependence on traveling by private cars and reduce or postpone a car purchase [1,2]; while some argue that shared mobility may be a compromise for individuals when they cannot afford cars, so shared mobility that appears in the transitional stage will prompt more individuals to buy private cars [3,4]. According to China statistics, from 2016 to 2020, the number of new car registrations shows a slow downward trend from 2752 million units in 2016 to 2424 million units in 2020 [5]. ...
Article
Full-text available
Online car-hailing services have become an integral part of people’s daily travel in China. Considering that young people are the major consumer group in car purchases, it is worth investigating how the experience of online car-hailing services affects their intention to purchase a car. Based on the extended theory of Planned Behavior, this study found that the factors that negatively impact the car purchase intention of the youth are firstly the public transportation service quality, followed by the risks of private cars, and finally the online car-hailing services quality. Elevating the convenience and comfort of public transportation is conducive to reducing car purchasing intention. The indirect effect of online car-hailing services on car purchase intention is greater than its direct effect, and the most important factor is attitude. The car purchase intention is significantly heterogeneous across age and annual household income groups. Improving the convenience of public transportation will reduce the car purchase intention of people in the early youth. For middle and later youth, providing demand-responsive transit for important individuals to meet their diverse needs can reduce car purchase intention. As for online car-hailing services, youth care most about their convenience and comfort and worry most about their safety. Providing better online car-hailing services can reduce the car-purchasing intention of youth.
... More recently though, studies have benefited from rich datasets of carsharing including pilot data (Hui et al., 2017) and booking data in order to identify the factors of interest for carsharing demand (Schmöller et al., 2015;Müller et al., 2017). Factors af-50 fecting demand included weather conditions, time of the week, and socio-demographics. ...
Article
Scooter–sharing has recently emerged as the newest trend in shared–mobility and micro–mobility; electric standing scooters are seen on the streets of major cities and are perceived as a fun, convenient mode of transport. However, there are also concerns regarding scooter safety, riding, and parking regulations. A motivation is to understand the impacts of scooters and their potential to disrupt existing systems. In this paper, the shift from carsharing to scooter–sharing is of particular interest. A stated preference survey targeting young individuals (18–34 years old) conducted in Munich was used to estimate a choice model between carsharing and scooter–sharing. The model was then applied to scenarios developed based on trip characteristics of a carsharing dataset. The model shift was then estimated for the scenarios, followed by a sensitivity analysis. In the best–case scenario, scooters were found to attract about 23% of carsharing demand.
... This is evidenced by numerous studies conducted in e.g. Poland (Tchorek, Brzozowski, Dziewanowska, Allen, Kozio , Kurtyka, & Targowski, 2020), China (Hui, Wang, Ding, & Liu, 2017;Yoon, Cherry, & Jones, 2017), Canada (Namazu & Dowlatabadi, 2018), Germany (Burghard & Dütschke, 2019) or in French corporations (Fleury, Tom, Jamet, & Colas-Maheux, 2017), as well as in the field of electric car sharing in Swe-den (Sopjani, Stier, Ritzén, Hesselgren, & Georén, 2019). ...
Article
Full-text available
The aim of the article is to present car sharing, with particular emphasis on electric car sharing, as an interdisciplinary research area. This applies not only to social sciences – management (strategy or marketing), sociology, economics (including the sharing economy), consumer psychology, but also to urban planning, engineering sciences (electrical engineering or energy) and, finally, ecology. Only the use of a broader perspective allows the understanding of the importance of car sharing, including electric vehicles, in contemporary social and economic processes. The diagnosis of factors that may affect the widespread use of car sharing, which we treat as an element of cities’ response to congestion and smog, requires a reference to the knowledge of the previously mentioned scientific disciplines. The core value of this article is that it provides a multi-faceted perspective on the consumer and prosumer, urban mobility and the energy ecosystem from the point of view of the sharing economy and zero/low carbon cars. In recent years, the number of research articles on car sharing has been growing (Ferrero, Perboli, Rosano, & Vesco, 2018); however, studies written from the point of view of a single, less often two scientific disciplines dominate. We propose to extend this perspective. Although, in research terms, this work is preliminary and exploratory, adopting a broad observation perspective should allow for establishing a dialogue between disciplines to ensure better formulation of research problems and solve socio economic dilemmas not only in the field of the sharing economy, and to better introduce the issue of car sharing to the area of management sciences.
... More recently though, studies have benefited from rich datasets of carsharing including pilot data (Hui et al., 2017) and booking data in order to identify the factors of interest for carsharing demand (Schmöller et al., 2015;Müller et al., 2017). Factors af-50 fecting demand included weather conditions, time of the week, and socio-demographics. ...
Article
Full-text available
Problematyka badania obejmuje rozwój mobilności współdzielonej w Polsce. Analizie poddano zarówno determinanty rozwoju polityk mobilnościowych w kraju, jak i za granicą. Wyznaczono również bariery oraz tendencje rozwojowe. W opracowaniu wykorzystano następujące metody badawcze: przegląd i krytyczną analizę literatury przedmiotu i opracowań branżowych, analizę statystyczną oraz analizy GIS. Przeprowadzone analizy wskazują jednoznacznie na to, że transport współdzielony systematycznie zyskuje popularność. Uwarunkowane jest to m.in. zmianą podejścia decydentów do rozwoju ośrodków miejskich, obierając w swoich działaniach rozwojowych bardziej zrównoważony kierunek. Transformacja ta zachodzi również w podejściu do rozwoju przestrzenno-funkcjonalnego miast. Obejmuje ona odejście od tradycyjnego modelu projektowania z nastawieniem na użytkowanie samochodów do zwrócenia większej uwagi na potrzeby mieszkańców. Wynika to również ze zmiany postaw i świadomości użytkowników miejskich, którzy dostrzegają negatywny wpływ środowiskowy, wynikający z ich preferencji transportowych. Opracowano i poddano analizie zestawienia statystyczne w krajach Europy Środkowo-Wschodniej. Wyniki wskazują na to, że Polska jest liderem w zakresie wolumenu zainteresowania użytkowników korzystaniem z transportu współdzielonego. Wykorzystano również studium przypadku Trójmiasta do zbadania atrakcyjności usług transportu współdzielonego na jego terenie. Wyznaczono zarówno obszary koncentracji usług współdzielonych, jak i obszary, w których transport publiczny jest niedostatecznie rozwinięty, a mobilność współdzielona może tę lukę uzupełnić.
Article
Sharing seems a key feature of transforming linear consumption to a more environmentally friendly system. This is especially applicable to car sharing. The aim of this study is to find out which factors influence environmentally friendly behaviour and how strongly. 13,629 journeys of a German car sharing provider specialised in the transport of goods and larger groups of people are evaluated. The focus is on the possibility for customers to offset their carbon footprint by voluntarily making their journeys climate neutral. Considering socio-economic characteristics, a Light Gradient Boosting Machine (LightGBM) model is applied to analyse variables which influence environmentally friendly behaviour. Age, place of residence, mileage driven, and education level have a statistically significant influence in predicting whether a customer will voluntarily offset CO2 or not, in contrast to gender. These findings have societal and political implications which could be used for future policy making.
Article
Abstract This study explores the mobility patterns of carsharing members from their trip distance perspective and its associated factors with a specific focus on its members' personal, usage, and stations' locational characteristics. Using Seoul as a case study, one-month rental transaction datasets provided by two-way carsharing operators were used as a data source. The multilevel mixed-effect modeling approach was applied to remedy spatial heterogeneity across station locations that affect the distance traveled by each rental. In addition, a classification among the carsharing members based on trip distance was conducted using regression tree to obtain clusters of the most homogenous member groups. The multilevel model results confirmed the important roles played by the station location and individual-level factors that affect mobility patterns of carsharing members. Individual-level characteristics showed that members in their 50s and female travel longer. Similarly, rentals made on non-workdays and in the morning showed longer travel distances. The station-level characteristics indicate that carsharing stations' proximity to public transit and leisure areas positively affects trip distances, suggesting the effect of the built environment and land use on the travel patterns of carsharing members. By combining carsharing transaction and their stations’ built environment data, this study suggests a new interface for city officials and carsharing operators to work together for achieving their sustainable mobility objectives together.
Article
Carsharing is an essential part of the transformation towards sustainable mobility in smaller urban areas. To expand their services and the positive social and environmental benefits, carsharing operators must understand their users' travel behavior. To accelerate this understanding, we analyze usage data of a station-based carsharing service from a small city in Germany with machine learning and explainable artificial intelligence to reveal influencing factors on the trip distance. The resulting four overarching groups are personal characteristics, time-related, car-related, and environmental features. We further analyze the driving distance of several subgroups split by personal and time-related features. Our findings highlight the importance of time-related features for the trip distance of carsharing users in all subgroups. We also discuss the influence of non-time-related features on the user groups. With these results, we derive valuable insights for research and carsharing operators by understanding patterns in individual user behavior in smaller urban areas.
Article
Sustainable development of carsharing plays an important role in reducing the number of private cars and relieving urban parking pressure. Understanding the behavior of users, such as spatiotemporal characteristics of carsharing users, can help operators find high-quality users and provide strong decision support for their service management. To this end, user loyalty based on user behavior is analyzed in this study. Moreover, station distribution of different loyalty types of users is revealed, which is beneficial to optimize the resource distribution. Deep belief network is constructed to predict the future loyalty of users based on order data. An improved two-stage clustering method is designed to mine station usage characteristics of users and determine activity areas of various users. Results show that prediction accuracy of user categories can reach 85% during the six-month observation period. The method of this study can provide a theoretical foundation for the development of a user-centric operation for carsharing.
Article
Carsharing is just a small niche product on the mobility market. The proven positive effects of carsharing can contribute to solving traffic problems only if the number of carsharing customers grows. A household survey was used to investigate the awareness and the market potential of mobility service carsharing in Germany. The results show that the majority of the respondents do not know what the term “carsharing” means. Even within the subgroup of people who could explain carsharing, local carsharing offers are not well known. To measure the attitudes toward the different modes of transport and acceptance of the general idea to share a car with other people, statement batteries were used. On the basis of factor analyses with linear and logistic regression models, the factors are determined that influence whether a person likes carsharing. Furthermore, the correlation between attitudes and behavioral aspects is revealed. In this context, people with multimodal mobility behavior are found to be more open-minded for shared-use vehicle systems. Finally, by taking subjective (attitudes) and objective (current mobility behavior) criteria into account, the potential of carsharing is estimated.
Article
This paper presents an econometric model for the behaviour of carsharing users. The econometric model is developed to jointly forecast membership duration, the decision to become an active member in a particular month, and the frequency of monthly usage of active members. The model is estimated using the membership directory and monthly transaction data of a carsharing program, 'Communauto Inc.', based in Montreal, Canada. The model shows a high degree of fit to the observed dataset and provides many behavioural details of carsharing users. The results are instructive to carsharing planners in devising efficient policies.
Article
Carsharing systems are gaining new members every month. However, few studies formally define the system and illustrate the systematic processing of administrative data sets to estimate indicators regarding both demand and supply objects. The first outcome of this research is the definition of the object model for a carsharing system. Rich transaction data sets, generally used for the production of monthly bills, are used to estimate indicators describing how the carsharing system is used in the Montreal area of Quebec, Canada. Indicators describing the main objects of the system-members, trip chains (transactions), cars, and stations-are estimated by using continuous data. The demand object analysis focuses on the study of members and their trip chains using the shared cars. The analysis shows that carsharing members are younger than the overall population, with an overrepresentation of 25- to 39-year-olds. The persistency of active members within the carsharing system is estimated at around 60% after 4 months and 50% after 12 months. The supply-objects analysis focuses on the study of cars and stations. Spatial dispersion of members with respect to stations used and typical use of cars is illustrated over long periods. With the increase in the number of members, transactions, cars, and stations, carsharing organizations need to find new ways to manage growth and optimize their networks. A clearer understanding of how their systems are used will help them develop enhanced planning and modeling abilities.
Article
A tool to assess the market potential for new carsharing operations in urban communities is examined and applied. The research is based on the analysis conducted for TCRP Report 108: Carsharing: Where and How It Succeeds. Geographic market segments in urban areas are analyzed. A geographic information system (GIS)–based analysis of 13 U.S. regions finds that neighborhood and transportation characteristics are more important indicators for carsharing success than the individual demographics of carsharing members. Results indicate that low vehicle ownership has the strongest, most consistent correlation to the amount of carsharing service in a neighborhood. Thresholds based on analysis results are outlined for low service (i.e., carsharing may be viable but limited growth can be expected) and high service (i.e., carsharing is likely to flourish). This tool to identify neighborhoods that can support carsharing is applied to a community seeking to establish a carsharing program: Austin, Texas. The analysis finds that several Austin neighborhoods have the characteristics to support carsharing (e.g., low vehicle ownership rates and high percentages of one-person households), but few Austin neighborhoods could support a high level of carsharing service.
Article
TRB’s Transit Cooperative Research Program (TCRP) Report 108: Car-Sharing--Where and How It Succeeds examines development and implementation of car-sharing services. Issues addressed in the report include the roles of car-sharing in enhancing mobility as part of the transportation system; the characteristics of car-sharing members and neighborhoods where car-sharing has been established; and the environmental, economic, and social impacts of car-sharing. The report also focuses on car-sharing promotional efforts, barriers to car-sharing and ways to mitigate these barriers, and procurement methods and evaluation techniques for achieving car-sharing goals. Appendices A through E of TCRP Report 108 are included with the report on CRP-CD-60 that is packaged with the report. The appendices include an annotated bibliography; a list of partner organizations surveyed and interviewed; survey instruments; and sample documents such as Requests for Proposals (RFPs) and zoning ordinances related to car-sharing. Appendix E was designed as a resource for introducing organizations to car-sharing and encouraging partnerships to initiate car-sharing programs.
Article
In land-scarce Singapore, cars are largely unaffordable due to the imposition of restraints such as high taxes and vehicle quotas. The car-sharing concept was introduced to satisfy citizens’ aspiration to use a car. The paper discusses the operational aspects of the car-sharing co-operative and analyses findings from surveys of 45 members and 76 non-members of the co-operative. The profile of members revealed a mean household size of 4.1, 73% of married status, 49% from middle-income households and 62% having never owned a car before. Members still mainly used public transport for travelling to work after attaining membership but turned more often to the co-operative car rather than public transport for marginal uses such as leisure and social trips.
Article
So far, car sharing is just a small niche product on the mobility market. The proven positive effects of car sharing can only contribute to solving traffic problems if the number of car sharing customers grows. The present paper investigates, on the basis of a household survey, the awareness and the market potential of mobility service car sharing in Germany. The results show that the majority of the respondents do not know what the term car sharing stands for. Even within the sub-group of people who could explain what car sharing is, local car sharing offers are not well known. To measure the attitudes towards the different modes of transport and the acceptance of the general idea to share a car with other people, statement batteries were used. On the basis of factor analyses, linear and logistic regression models, the factors are determined that influence whether a person has a liking for car sharing or not. Furthermore, the correlation between attitudes and behavioral aspects are revealed. In this context, people with multimodal mobility behavior are found to be more open-minded for shared used vehicle systems. Finally, by taking subjective (attitudes) and objective criteria (current mobility behavior) into account, the potential of car sharing is estimated. The paper starts with a short history of the development and the current status quo of car sharing in Germany and a brief summary of the previous research. In the end recommendations for the further development of car sharing are given.