Research on the Application and Management of Shared Bikes in

Smart Cities

Weiqing Liu

EDP School of Intelligent Engineering, Hubei Enshi College, Enshi, Hubei, 45000, China

Keywords: Shared Bike Management, Smart Cities, Urban Air Quality, Sustainable Transportation.

Abstract: As a typical means of shared transportation, the application and management of shared bikes face some

challenges. Therefore, the application and management of shared bikes in smart cities have been studied

indepth in this paper. This study founds that an effective management system and policies are needed to ensure

the healthy development of the bike-sharing industry. In addition, the relationship between bike-sharing and

urban air quality. The use of a large number of motor vehicles has led to increasingly serious air pollution

problems, and shared bikes, as a low-carbon and environmentally friendly mode of transportation, can reduce

vehicle exhaust emissions and improve air quality. In summary, Shared cycling in the wisdom of city

application and management is facing some problems, but through the optimization measures, intelligent

control methods and air quality improvement strategy related research, can effectively solve these problems,

the government, Shared cycling enterprises and related departments of cooperation and regulation is a key

factor in the management of Shared bicycle therefore, Shared cycling managers and policy makers need to

pay attention to these research results, formulate corresponding policies and measures to promote the

sustainable development of Shared cycling industry.

1 INTRODUCTION

With the rapid growth of the urban population and the

aggravation of traffic congestion problems, sharing

traffic and emerging transportation modes have

become one of the important means to reduce the

traffic pressure. The development of shared

transportation and emerging transportation modes in

smart cities is becoming a focus of research. As a low-

carbon, environmentally friendly and convenient way

of transportation, shared bikes have gradually

attracted people's attention. However, the chaos

caused by the excessive number of shared bikes or

mismanagement has brought considerable problems

to city managers. Therefore, it is necessary for the

application and management of shared bikes in smart

cities.

In recent years, many studies have deeply

explored the application and management of shared

bikes. By analyzing the use data of shared bikes,

Zhang et al. proposed some optimization measures,

such as reasonably adjusting the distribution of

vehicles and setting up parking spots, to reduce traffic

congestion and improve user experience (Lei et al

2017). Zhang discusses the intelligent control method

to realize the shared bikes in the smart city

environment class human beings (Wang et al 2018).

Burda Studying the strategy of implementing public

bike sharing in Dhaka City, Bangladesh (Zhang and

Zhou 2019). Shaheen et al. studied early operator

understanding and trends of public bike sharing in

North America (Burda and Hakim 2017). Sun studied

the relationship between shared bikes and urban air

quality by analyzing the use of shared bikes and urban

air quality data and proposed some strategies to

improve air quality, such as encouraging the use of

shared bikes instead of motor vehicles (Shaheen et al

2015). To sum up, sharing transportation and

emerging transportation modes are of great

significance in smart cities. As one of its

representatives, shared bikes have management

problems, but reasonable management and technical

means can give full play to their advantages and bring

convenience and environmental benefits to urban

transportation. Therefore, it is of great practical

significance to strengthen the research and

management of shared transportation and emerging

transportation modes.

This paper aims to make a comprehensive

analysis of the application and management of shared

234

Liu, W.

Research on the Application and Management of Shared Bikes in Smart Cities.

DOI: 10.5220/0012815600004547

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

In Proceedings of the 1st International Conference on Data Science and Engineering (ICDSE 2024), pages 234-241

ISBN: 978-989-758-690-3

bikes in smart cities. This paper adopts the scientific

data collection method, collects a large number of

data, including user data, vehicle data and traffic data,

and uses the statistical principle to make a

comprehensive analysis and processing of the data.

Finally, this paper makes an in-depth analysis of the

cases of shared bike management in different cities

and different enterprises and summarizes the

problems and solutions existing in the management of

shared bikes. The results of this paper can provide

valuable reference and suggestions for bike-sharing

managers and policy makers, and promote the healthy

development of the bike-sharing industry.

2 METHODS AND DATA

2.1 Data Source

The data source of this article is the Washington, D.

C. Bike sharing system is a way of renting bikes, the

process of automatic membership, renting and

returning bikes through the network of kiosks

locations throughout the city. Using these systems,

people can rent a bike from one place and return it

elsewhere as needed. Currently, there are more than

500 bike-sharing programs worldwide. The data

generated by these systems make them attractive to

researchers by explicitly recording the duration of

travel, departure location, arrival location, and

elapsed time. Thus, bicycle-sharing systems serve as

sensor networks that can be used to study urban

mobility. Combine historical usage patterns with

weather data to predict bike rental demand in the

Capital Bike Sharing program in Washington, DC.

2.2 Application of Time Series Analysis

in Bike-Sharing Research

Time series analysis is an important statistical method

that can dig deep into patterns and trends over time

from data points. In the field of bike-sharing research,

time series analysis plays a key role, helping us to

fully understand the dynamic trends and patterns of

the use of shared bikes.

Through sophisticated time series analysis, this

paper can predict the amount of shared bikes used in

different periods (for example, different time periods

of the day, different days of the week, etc.). This

predictive ability is crucial to the effective

management and scheduling of bike-sharing

companies. It can help companies better plan vehicle

distributions and optimize scheduling algorithms to

meet user needs and improve operational efficiency.

Furthermore, time series analysis can also reveal

seasonal and cyclical changes in bike-sharing use.

These findings help companies better understand user

needs and develop targeted operational strategies.

Moreover, time series analysis can also detect

abnormal usage behavior, such as a sudden increase

in usage in a short period or a prolonged sustained

trough. These abnormalities may mean the

occurrence of certain special events, such as bad

weather, holidays, large events, etc. With this

information, companies can adjust their operational

strategies promptly to address possible challenges.

2.3 Selection and Suitability of the

ARIMA Model

ARIMA model was used for time series prediction.

The model is particularly suitable for non-stationary

time series data and enables efficient analysis of the

changes in shared bike usage over time. By studying

past data trends and patterns, the ARIMA model

predicts the use trend in the future and provides a

scientific basis for the effective management of

shared bikes.

To be more accurate, this study also involves the

analysis of autocorrelation and partial

autocorrelations. These analyses help to determine

the parameters in the ARIMA model, such as lag

order (lag) and differential order. By analyzing the

autocorrelation and partial autocorrelation of shared

bike use data, this paper can have a deeper

understanding of the change law of bike use over

time, to predict the future use trend more accurately.

3 TIME SERIES ANALYSIS OF

SHARED BIKE USAGE DATA

3.1 Analysis of Autocorrelation and

Bias Toward Autocorrelation

In this paper, SPSS 25 and ARIMA models were used

for time series analysis and prediction in hours. The

analysis of autocorrelation and partial autocorrelation

revealed sequences with significant and partial

autocorrelation, especially in a 1 to 6 h delay (Figure

1). This suggests that the past values of these

sequences have significant effects on the future

values and that this effect diminishes over time. This

analysis is important for understanding changes in

time series data and for predicting future trends.

Research on the Application and Management of Shared Bikes in Smart Cities

235

Figure 1: Time series analysis (Picture credit: Original).

Autocorrelation (ACF) and partial autocorrelation

(PACF) are commonly used in time series analysis. A

sequence autocorrelation diagram is a (linear)

diagram between a sequence and its lag. Figure 2

shows the hysteresis weights as a function of

hysteresis. How the correlation between individual

time steps decrease or disappear with increasing time

steps. Sequence autocorrelation graph is a method to

test the presence of sequences.

Figure 2: The hysteresis weights (Picture credit: Original).

The partial autocorrelation coefficient plot shows

the partial autocorrelation coefficient as a function of

hysteresis. The partial autocorrelation coefficient can

be seen as the remaining effect after removing some

of the effects that have been explained by the previous

lag value. Thus, the partial autocorrelation coefficient

plot can be used to determine the number of periods

p in the autoregressive AR model. The p-value of this

model is 1.

This indicates that there is a time series

association of shared bike usage, that is, there is a

mutual correlation between the previous period and

the latter period.

3.2 Non-Simulated Sequence and the

ARIMA Model

The instability of the time series means that the

statistical characteristics of the data such as the mean

and variance will change significantly at different

time points. Therefore, the overall characteristics of

the time series cannot be described by a fixed model,

but the model needs to be adjusted according to the

time change. The effective analytical method for non-

stationary sequences is the ARIMA (1,0,1) model.

This model captures the dynamic changes of the time

series by combining the differential and moving

average terms and can fit the non-stationary time

series data relatively well. Where "1" indicates the

order of the difference, "0" indicates the order of the

autoregressive part, and "1" indicates the order of the

moving average part. By tuning these parameters, the

ARIMA model can adapt to different time-series data

features (Table 1). The purpose of this study is to use

the ARIMA (1,0,1) model to conduct an in-depth

analysis of non-stationary time series data, to better

understand the internal laws and trends of time series

data and provide strong support for subsequent data

analysis and prediction.

As shown in Table 2, the parameter display,

autoregressive (AR), and moving average (MA)

terms of the ARIMA model all have significant

effects on the model, which reveals the complexity

and dynamics of the usage patterns of shared bikes.

Table 1: Model description.

types of models

model ID count model _1 ARIMA(1,0,1)(0,0,0)

Table 2: Model fit.

Fitting statistics average value standard error least value crest value

centile

5 10 25 50 75 90 95

Stable R square .764 . .764 .764 .764 .764 .764 .764 .764 .764 .764

R square .764 . .764 .764 .764 .764 .764 .764 .764 .764 .764

ICDSE 2024 - International Conference on Data Science and Engineering

236

3.3 Model Size and Prediction Accuracy

After the statistics of the output degree of the model,

a series of indicators, such as stationary R square,

RMSE and MAPE, are obtained(Table 3). They

provide an important basis for us to evaluate the

accuracy and stability of the selected model in the

actual data prediction. From these statistics, this paper

can see the validity of the model in predicting the use

of shared bikes.

Specifically, the stationary R square value is

0.764, which is quite high, meaning that the model

can explain 76.4% of the variation in the real data.

This result fully demonstrates the powerful ability

and accuracy of the model in capturing and predicting

the usage of shared bikes, indicating that the model

can effectively apply the patterns and trends in

historical data to predict the future usage of shared

bikes.

In addition, other indicators such as RMSE (root

mean squared error) and MAPE (average absolute

percentage error) also further confirm the superiority

of the selected model in the output of actual data.

These statistics provide us with a more

comprehensive perspective, allowing us to more

accurately evaluate the predictive performance of the

model.

In conclusion, based on the output degree

statistics of the model, this paper can conclude that

the selected model is valid and reliable in predicting

the usage of shared bikes. This conclusion provides

strong support for decision-makers to enable them to

make rational decision making and planning based on

these prediction data.RMSE and MAPE values were

87.928 and 164.601, and these statistical measures

further verified the accuracy of the model predictions.

Figure 3 shows the fitting and measured results,

which show that the effect is better.

Figure 3: The fitting results (Picture credit: Original).

4 DESCRIPTIVE STATISTICS

AND REGRESSION ANALYSIS

4.1 Descriptive Statistical Analysis

In this analysis, this paper used the correlation

coefficient to quantify the relationship between the

analyzed terms (Table 4). First, by calculating the

correlation coefficient, this paper assessed whether

these relationships were present. Then, by positive

and negative symbols, this paper determined the

direction of these relationships. Moreover, the

magnitude of the correlation coefficient also reflects

the strength of the linear relationship between the

variables.

The Pearson correlation coefficient (PCC)and the

Spearman correlation coefficient are two commonly

used correlation coefficients, both of which can be

used to describe the degree of correlation. The basic

criteria for these two coefficients are consistent.

Generally, when the absolute value of the correlation

coefficient is greater than 0.7, this paper can that a

strong correlation is greater than 0.4, and when the

absolute value is less than 0.2.

In the previous section, this paper performed a

descriptive statistical analysis of the sample data and

found that the data used had a certain degree of rigor

and rationality. Next, this paper will initially judge

the degree of correlation between the variables

through the correlation analysis.

Table 3: Model statistics.

model Number of

predictive

variables

Model fit degree statistics Young-Box Q (18) The

number of

outliers

Stable R

square

R square normalization BIC statistics DF conspicuousness

count-

model _1

0 .764 .764 8.956 2287.039 16 .000 0

Research on the Application and Management of Shared Bikes in Smart Cities

237

Table 4: Descriptive statistics.

least value crest value mean standard deviations

season 10886 1 4 2.51 1.116

holida

10886 0 1 .03 .167

workin

10886 0 1 .68 .466

weathe

10886 1 4 1.42 .634

temp 10886 .82 41.00 20.2309 7.79159

atemp 10886 .760 45.455 23.65508 8.474601

humidit

10886 0 100 61.89 19.245

windspee

10886 .0000 56.9969 12.799395 8.1645373

casual 10886 0 367 36.02 49.960

istere

10886 0 886 155.55 151.039

coun

10886 1 977 191.57 181.144

umber of valid cases

(

in a column

)

10886

Table 5: Relativity.

season holiday workingday weather temp atemp humidity windspeed casual registered count

season PCC 1 .029* -.008 .009 .259* .265* .191* -.147* .097* .164* .163*

holiday PCC .029* 1 -.250* -.007 .000 -.005 .002 .008 .044* -.021* -.005

workin

PCC -.008 -.250* 1 .034* .030* .025* -.011 .013 -.319* .119* .012

weather

PCC .009 -.007 .034* 1 -.055* -.055* .406* .007 -.136* -.109* -.129*

temp

PCC .259* .000 .030* -.055* 1 .985* -.065* -.018 .467* .319* .394*

atemp

PCC .265* -.005 .025* -.055* .985* 1 -.044* -.057* .462* .315* .390*

humidit

PCC .191* .002 -.011 .406* -.065* -.044* 1 -.319* -.348* -.265* -.317*

winds

PCC -.147* .008 .013 .007 -.018 -.057* -.319* 1 .092* .091* .101*

casual

PCC .097* .044* -.319* -.136* .467* .462* -.348* .092* 1 .497* .690*

registered

PCC .164* -.021* .119* -.109* .319* .315* -.265* .091* .497* 1 .971*

count

PCC .163* -.005 .012 -.129* .394* .390* -.317* .101* .690* .971* 1

*. At the 0.01 level (two-tailed), the correlation was significant.

*. At the 0.05 level (two-tailed), the correlation was significant.

Table 6: Model summary.

model R R s

uare Ad

uste

R s

uare Error in the standard estimation

1 .743

.552 .551 121.325

a. Predictor variables: (constants), casual, holiday, windspeed, weather, season, workingday, attemp, humidity, temp

In this section, the Pearson coefficient and Spearman

coefficient were used for correlation. Among them,

the Pearson coefficient was used in this analysis.

After testing, the results are shown in the Table 5.

According to the descriptive statistical analysis,

the use of shared bikes was significantly associated

with factors such as season, holidays, weekdays,

weather and temperature. Specifically, the season has

a particularly significant impact on the shared usage

of bikes, which reflects users' concern about the

weather conditions in different seasons.

4.2 Regression Analysis

The R side, also known as the determination

coefficient, is an indicator of quantifying the

goodness of fit of the model (Sun et al 2018). Its

explanatory ability is crucial because it can show the

degree to which the model interprets the data. The

closer the R square value is to 1, the higher the

goodness of fit of the model is. In this case, the R

square value is 0.551, indicating that the model is a

good fit (Table 6).

ICDSE 2024 - International Conference on Data Science and Engineering

238

Table 7: ANOVAa.

model

uadratic su

free de

ree mean s

uare F cons

icuousness

1 regression 197079673.678 9 21897741.520 1487.632 .000

residual 160093239.998 10876 14719.864

amount to 357172913.676 10885

a. Dependent variable: coun

. Predictor variables: (constants), casual, holiday, windspeed, weather, season, working day, attempt, humidity, temp

Table 8: Coefficient.

model

Unstandardized coefficients Standardization coefficient

t cons

icuousnessB Standard erro

Beta

1 (constant) 13.454 6.938

1.939

.053

season 17.453 1.113 .108 15.683 .000

holiday 23.985 7.230 .022 3.317 .001

workingday 95.916 2.829 .247 33.909 .000

weather -2.765 2.042 -.010 -1.354 .176

temp -1.622 .890 -.070 -1.821 .069

atemp 1.908 .819 .089 2.329 .020

humidity -.680 .077 -.072 -8.775 .000

windspeed .626 .156 .028 4.017 .000

casual 2.611 .031 .720 84.010 .000

a. Dependent variable: coun

As shown in Table 7, F (1487.632), significance

p(0.00) is less than 0.05, indicating that the

correlation coefficient of the regression equation is

not 0, indicating that the regression equation is

meaningful.

According to the regression coefficient in Table 8,

the significance of all the variables except weather

temp is less than 0.05, indicating that all the other

variables have a significant impact on the count.

Regression analysis revealed the extent of

different factors on the use of shared bikes. For

example, there are significant differences in the

effects of weekdays and holidays on usage, which

may be related to the travel habits and daily activity

patterns of urban residents.

4.3 Discussion and Management

Suggestions Combined with Data

Analysis

Data-driven management decision-making: The data

analysis results of this paper highlight the importance

of using data-driven decision-making methods when

implementing bike-sharing management in smart

cities. By analyzing the time series data of shared bike

use, city managers can more effectively predict and

respond to changes in the demand for shared bikes

(Sun et al 2019 & Li et al 2020).

Scheduling strategy for weather forecast:

regression analysis and correlation analysis reveal

key factors affecting the use of shared bikes, such as

weather, temperature, and urban activity. These

findings could guide city managers to develop

scheduling strategies for more weather forecasting,

such as adjusting the allocation of shared bikes under

specific weather conditions or holidays.

Standardized management system: Combined

with the prediction results of the ARIMA model, a

smart cities can develop a standardized bike-sharing

management system, which can adjust the allocation

of shared bikes in real time, optimize user experience,

and alleviate traffic congestion.

5 CHALLENGES AND

OPPORTUNITIES

Bike-sharing has played an important role in relieving

urban traffic pressure, but it also brings new

challenges. Due to the rapid growth of shared bikes,

some cities have experienced problems such as

excessive concentration and parking in disorder,

affecting the urban landscape and traffic order.

In addition, the peak use period of bike-sharing

often coincides with the commuting rush hour in the

city, which strengthens the traffic congestion in the

city to some extent.

In the context of smart cities, shared bikes

provide an environmentally friendly and fast way to

Research on the Application and Management of Shared Bikes in Smart Cities

239

make short trips, which helps to reduce urban carbon

emissions and improve residents' travel efficiency

(Zhang et al 2020).

The popularity of bike-sharing also provides

valuable data sources for data collection and analysis

in cities, helping city managers to better understand

urban traffic patterns and residents' travel needs.

In the future, the bike-sharing industry may be

further developed through technological innovation,

such as the use of more advanced positioning and

navigation technologies, to improve the utilization

rate and management efficiency of the bikes (Zhang

and Qi 2018).

With the development of Internet of Things

technology, shared bikes can be more customized,

such as real-time data analysis to automatically adjust

the allocation of bikes to better meet user needs.

To better manage and develop bike-sharing, city

managers need to develop reasonable strategies and

policies, such as optimizing the urban traffic layout,

providing more special bicycle lanes, and

establishing reasonable charging and weak

mechanisms.

At the same time, the government and enterprises

should strengthen cooperation to jointly promote the

sustainable development of shared bikes and provide

more convenient, efficient, and environmentally

friendly travel options to urban residents.

6 CONCLUSION

This paper studies the challenges and opportunities of

bike-sharing management through data analysis, and

puts forward corresponding management

suggestions. As a convenient and environmentally

friendly means of transportation, shared bike has

developed rapidly in cities in recent years. However,

with the increasing number of shared bikes, a series

of management problems have also emerged. This

paper first analyzes the imbalance between supply

and demand faced by shared bikes. Due to the uneven

distribution of shared bikes, there may be a surplus in

some areas, while others may have insufficient bikes.

In addition, shared bikes are also prone to excessive

aggregation of bikes, making it difficult for users to

find available bikes. To address these problems, this

paper proposes a data-driven management decision

strategy. Through the data analysis of user behavior

and demand, the demand change of shared bikes can

be predicted, and the allocation of vehicles can be

adjusted in advance to achieve the goal of supply and

demand balance. Secondly, this paper also studies the

impact of the weather on the use of shared bikes.

Weather is one of the most important factors for

people to choose their travel tools. For example, in

bad weather, people prefer to use public transport or

taxis rather than ride shared bikes. Therefore, this

paper believes that weather factors need to be taken

into account when formulating scheduling strategies

to better allocate shared bikes and improve their

utilization rate and management efficiency.

Finally, this paper puts forward some suggestions

for the management of shared bikes. First of all, a

standardized bike-sharing management system

should be established. Through unified management

and scheduling, the balanced distribution of bikes can

be realized and the user experience can be optimized.

Secondly, the government can formulate

corresponding strategies and policies to support the

development of shared bikes, such as optimizing the

transportation layout, establishing special bicycle

lanes, and formulating charging and weak

mechanisms. For future studies, this paper suggests

that the sustainability of shared bike management can

be further explored. For example, how to achieve a

balanced allocation of shared bikes, improve

utilization and management efficiency, and how to

optimize the user experience. In addition, the

coordinated development of shared bikes and other

vehicles, and the impact of shared bikes on the urban

environment and traffic conditions can also be

studied. The significance of these studies is to provide

a scientific basis for the management of shared bikes

and promote the intelligent and sustainable

development of urban traffic.

In conclusion, this paper studies the challenges

and opportunities of bike-sharing management.

Further research can provide more profound

theoretical and practical guidance for the

management of bike-sharing, and promote the

sustainable development of the bike-sharing industry.

Through continuous exploration and innovation,

bike-sharing can truly become an important part of

urban transportation and provide residents with more

convenient, efficient, and environmentally friendly

travel options.

REFERENCES

Z. Lei, T. Liu, L. Yang, et al. Urban traffic research based

on shared bicycle data. Urban Transportation, 14(2),

123-132, (2017).

H. Wang, T. Xu, X. Jia, & W. Li. Shared bicycle

management in a smart city: A case study of

Hangzhou, China. Transportation Research Part C:

Emerging Technologies, 90, 58-71,. (2018).

ICDSE 2024 - International Conference on Data Science and Engineering

240

Y. Zhang & J. Zhou. Intelligent control of shared bike in

smart city environment. Communications in Computer

and Information Science, 1048, 287-296,. 2019.

J. M. Burda & S. Hakim. Shared bikes and the city:

Strategies for implementing public bike-sharing in a

city of Dhaka. IATSS Research, 41(2), 102-109,. 2017.

S. Shaheen, E. Martin, N. Chan, A. Cohen, & J. M.

Pogodzinski.. Public bikesharing in North America:

early operator understanding and emerging trends.

Transportation Research Record: Journal of the

Transportation Research Board, 2534(1), 78-87, 2015.

L. Sun, Y. Zhang, X. Wang, et al.. Research on the

relationship between shared bicycles and urban air

quality. Transportation and Transportation Research,

17(4), 65-71, (2018).

L. Sun, Y. Chen, C. Liu, & P. Yang.. Optimizing docking

station allocation in bike-sharing systems. IEEE

Transactions on Intelligent Transportation Systems,

20(10), 3950-3961, (2019).

M. Li, K. Chen, T. Wang, et al. Research on the layout

model of shared bicycle parking areas. Transportation

Planning and Management, 39(5), 94-100,(2020). .

J. Zhang, Z. Gui, S. Wang, & Y. Xu.The Application of

Shared Bicycles in Smart Cities. In International

Conference on Data Mining and Big Data (pp. 776-

783). Springer, Singapore. (2020).

Y. Zhang & S. Qi. A Research on Management of Shared

Bike in Smart City Based on Internet of Things. In

2018 13th International Conference on Computer

Science & Education (ICCSE) (pp. 897-902).

IEEE.(2018).

Research on the Application and Management of Shared Bikes in Smart Cities

241