Study on Influencing Factors of Population Changes in Dongguan

City Based on Principal Component-Regression Analysis

Zhitao Cui

1, a

, Shanle Wan

2, b

and Yong Fang

1, c

School of Artificial Intelligence, Dongguan City University, Dongguan, China

Dongguan Yeyuying Network Science Co. Ltd, Dongguan, China

Keywords: Influencing Factors, SPSS, Pearson Correlation Coefficient, Hierarchical Clustering, Principal Component

Analysis, Regression Analysis.

Abstract: In this paper, 9 economic and social development indicators are selected from the Statistical Yearbook 2019

of Dongguan. All indicator variables have a significant linear correlation with the permanent population

variable. The similarity of variables is distinguished with the hierarchical clustering method. Variable groups

are divided into 2 categories and then subjected to dimensionality reduction based on principal component

analysis. Two principal components are extracted. Finally, two regression equations are set up according to

the results of variable group clustering and principal component analysis, which have good test indicators.

1 INTRODUCTION

Since the reform and opening up, Dongguan's

economy has developed rapidly with its growth rate

and order of magnitudes among the top of major

economic cities in Guangdong Province. The

demographic dividend is one of the most important

factors (Cai, 2009; Zhao, 2016; Chen, 2008; Wang,

2007; Zhong, 2009). However, China's natural

population growth has entered a low level state since

the 1990s. By 2000, the population aged 65 and over

accounted for 7% in China, marking its entry into an

aging society (Yan, 2021). The decline in the growth

rate of the total supply of labor force combined with

the aging population structure will continue to affect

the innovation momentum and the potential growth

rate of the medium and long-term economy (Li,

2018). The formulation of policies conducive to

increasing population supply and slowing down the

aging process is an urgent task facing all major cities

in China.

The importance of labor supply to the

manufacturing industry is self-evident as Dongguan

is an international manufacturing city. It must work

out relevant policies conducive to promoting the

supply of labor force and slowing down the aging of

population in order to maintain the continuous

increase in social innovation momentum and the

continuous and stable development of the social

economy. To play a positive role in finding an

effective policy focus (Pan, 2012), this paper attempts

to identify the factors with an important impact on

population growth from the economic and social

development indicators of Dongguan.

2 CORRELATION ANALYSIS OF

DATA INDICATORS

The 9 data indicators herein are selected from the

Statistical Yearbook 2019 of Dongguan City, as

shown in Table 1.

Table 1: Interpretation of data indicators.

Variables Variable inter

retation

Regional Gross Domestic Product (GDP), in

100 million

Annual fixed asset investment

Number of large-scale industrial enterprises

Highway mileage

Number of teachers in regular institutions of

her learnin

Number of teachers in other types of schools

Number of students in regular institutions of

her learnin

Registered population

Natural population growth rate

Permanent population

Cui, Z., Wan, S. and Fang, Y.

Study on Inﬂuencing Factors of Population Changes in Dongguan City Based on Principal Component-Regression Analysis.

DOI: 10.5220/0011732800003607

In Proceedings of the 1st International Conference on Public Management, Digital Economy and Internet Technology (ICPDI 2022), pages 191-196

ISBN: 978-989-758-620-0

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

191

Table 2: Pearson correlation coefficients between permanent population and 9 data variables and their significant test results.

Pearson correlation

coefficient

.831

.833

.821

.881

.748

.849

.696

.872

-.488

.000 .000 .000 .000 .000 .000 .000 .000 .007

Figure 1: Cluster analysis of 9 data variables.

SPSS25.0 software is used to calculate the Pearson

correlation coefficients between the permanent

population and the 9 data variables (Deng, 2017) and

analyze influencing factors quantitatively (Wang,

2020), as shown in Table 2.

There is a significant linear correlation between

the permanent population and the 9 data variables at

the significance level. The number of teachers and

students in regular institutions of higher learning and

the natural population growth rate are moderately

related to the permanent population, while the other 6

variables are highly linearly related to the latter.

3 DISTINGGUISHING THE

SIMILARITY OF VARIABLE

GROUPS

R-type clustering is made in the cluster analysis, and

variables are classified based on their similarity.

Before clustering, the data must be standardized to

overcome the influence of dimensions (Zhao, 2014):

jij

−

(1)



ijj

(2)



−

jijj

)(

μσ

(3)

The clustering dendrogram is shown in Figure 1.

It is appropriate to divide variable groups into 3

categories by the cutting level of clustering (Wang,

1998) according to the clustering dendrogram.

Category I: Regional GDP, annual fixed asset

investment, highway mileage, number of teachers in

other types of schools and registered population; the

variables in this category mainly involve economic

development;

Category II: Number of large-scale industrial

enterprises, number of teachers in regular institutions

of higher learning and number of students in regular

institutions of higher learning; large-scale industrial

enterprises and institutions of higher learning are

similar in the agglomeration of floating population.

Category III: Natural population growth rate.

In order to have a definite understanding of the

clustering results, SPSS25.0 is used to give the

Pearson correlation coefficients between the 9 data

variables, as shown in Table 3.

ICPDI 2022 - International Conference on Public Management, Digital Economy and Internet Technology

192

The variable groups can be divided into 2 categories

by their Pearson correlation coefficients;

Category I: Natural population growth rate;

Category II: Economic and social development

indicators represented by GDP;(**Significant at

0.01).

4 DIMENSIONALITY

REDUCTION OF VARIABLE

GROUPS

In principal component analysis, the original

variables are transformed into a few new variables.

The sum of variances of the new variables is as

close as possible to that of the original ones.

Therefore, the number of variables is decreased,

achieving the effect of dimensionality reduction of

the data set (He, 2012). Information omission and

overlap are taken into account in the process.

The analysis results are shown in Table 4.

Table 4: Kmo test and Bartlett test.

KMO

value

Approximate

chi-square

Degree of

freedom

Sig

0.837 738.779 36 0.000

The KMO test value is 0.837 and Bartlett test

probability 𝑠𝑖𝑔 <0.05 . The data set is appropriate

for principal component analysis.

Table 5: Variance explanation of principal component

analysis.

Principal

component

facto

Eigenvalues Variance%

Cumulative

variance%

Component 1 7.658 85.084 85.084

Component 2 1.092 12.129 97.213

Two principal components are extracted and the

cumulative contribution rate of their variances is 97.213%.

Table 3: Pearson correlation coefficients between variable groups.

Pearson correlation

coefficient

Z-𝑥



Z-𝑥



Z-𝑥



Z-𝑥



Z-𝑥



Z-𝑥



Z-𝑥



Z-𝑥



Z-𝑥



Z-𝑥



0.992

∗∗

0.927

∗∗

0.961

∗∗

0.985

∗∗

0.998

∗∗

0.957

∗∗

0.982

∗∗

-0.056

Z-𝑥



0.951

∗∗

0.973

∗∗

0.968

∗∗

0.995

∗∗

0.931

∗∗

0.970

∗∗

-0.099

Z-𝑥



0.914

∗∗

0.899

∗∗

0.941

∗∗

0.843

∗∗

0.931

∗∗

-0.138

Z-𝑥



0.918

∗∗

0.968

∗∗

0.874

∗∗

0.943

∗∗

-0.197

Z-𝑥



0.977

∗∗

0.987

∗∗

0.963

∗∗

0.091

Z-𝑥



0.945

∗∗

0.983

∗∗

-0.096

Z-𝑥



0.927

∗∗

0.129

Z-𝑥



1 -0.085

Z-𝑥



Table 6: Factor loading matrix in principal component analysis.

Component matrix

Data indicators

Components

1 2

Regional GDP (RMB 100 million) .997 .014

Annual fixed asset investment (RMB 100 million) .995 -.037

umber of large-scale industrial enterprises (Nos.) .947 -.094

Highway mileage (km) .967 -.144

umber of teachers in regular institutions of higher learning over

the

ears

(

Nos.

)

.982 .165

umber of teachers in other t

es of schools

(

Nos.

)

.998 -.029

umber of students in regular institutions of higher learning (0,000) .952 .213

Registered population (0,000) .984 -.019

atural growth rate -.067 .993

Note: Extraction method: principal component

Study on Inﬂuencing Factors of Population Changes in Dongguan City Based on Principal Component-Regression Analysis

193

Table 7: Principal component coefficients.

1prin

2prin

0.3603 0.0134

0.3596 -0.0354

0.3422 -0.0900

0.3494 -0.1378

0.3549 0.1579

0.3606 -0.0278

0.3440 0.2038

0.3556 -0.0182

-0.0242 0.9503

Table 8: Test of regression coefficients between permanent population variable and regional GDP and natural population

growth rate in the standardized data.

Model Coefficient t Si

VIF

Regional GDP 0.807 12.444 0.000 1.003

Natural population growth rate -0.443 -6.835 0.000 1.003

In the first principal component, all variables

except the natural population growth rate have higher

values and equivalent orders of magnitudes,

indicating that the corresponding variables have

equivalently great influence on the first principal

component. Economic development has brought the

demand for population, while the agglomeration of

population has promoted economic growth (Liu,

2015; Wang, 2018; Gao, 2016). Population growth is

both the result and cause of economic development

(Smith, 2003).

In the second principal component, the natural

population growth rate has the greatest impact.

Calculation formula of principal components:

)4(042.0

3556.03440.00.36063549.0

3494.03422.03596.06033.01

xxxx

xxxxprin

×−

×+×+×+×+

×+×+×+×=

0182.02308.00.0278-1579.0

1378.0-09.0-0354.0-0134.02

xxxx

xxxxprin

×−×+××+

××××=

)5(9503.0

x×+

is the standardized variable of

𝑖  1...9.

5 REGRESSION ANALYSIS

Equations used in the regression analysis describe the

correlation between dependent variables and

explanatory variables. The process includes the least

squares estimation of coefficients of regression

equations, the significance test of equations and the

significance test of regression coefficients.

A regression model of the standardized permanent

population variable, standardized regional GDP

variable and standardized natural population growth

rate is established with the input method in the SPSS

regression analysis according to the results of the

cluster analysis of variable groups:

and . R

is adjusted

to 0.878.

Probability of significance test of the model

000.0=sig

Regression equation:

(6)

A linear regression equation is established for the

standardized permanent population variable and the 2

principal components:

ICPDI 2022 - International Conference on Public Management, Digital Economy and Internet Technology

194

Table 9: Test of regression coefficients between standardized permanent population variable and 2 principal components.

Model Coefficient t Si

VIF

𝑝𝑟𝑖𝑛1

0.303 13.452 0.000 1.000

𝑝𝑟𝑖𝑛2

-0.419 -7.014 0.000 1.000

𝑅 = 0.946 and 𝑅



= 0.895. R

is adjusted to 0.887.

Probability of significance test of the model 𝑠𝑖𝑔 =

0.000.

Regression equation:

2419.01303.0

prinpriny ×−×=

(7)

Substitute equations (4) and (5) into equation (7):

0.12090414.01636.0

1414.01238.00.1036

xxx

xxxy

×+×+×+

×+×+×=

4055.0-1541.00188.0 xxx ××+×+

(8)

It can be concluded that the 9 variables are

sequenced below based on the influence on the

permanent population variable:

xxxxxxxxx >>>>>>>>

(9)

6 DISCUSSION OF VARIABLES

IN THE REGRESSION MODEL

The regression model in this paper is explained as

below based on existing literature research results.

The labor-intensive enterprises represented by the

"three-plus-one" trading-mix and "three kinds of

foreign-funded enterprises" gathering in Dongguan in

the process of reform and opening-up have greatly

improved the demand for labor force and contributed

to the inflow of floating population (Gao, 1995).

Regional income differences represented by regional

GDP and the number of large-scale enterprises are

two manifestations of China's population migration

mechanism (Gao, 1995). From the perspective of

urban-rural dual economic structure, the

agglomeration of production factors and the

advantages of production methods in urban areas

have facilitated the flow of labor from the low-

productivity agricultural sector to the high-

productivity industrial sector (Kong, 2001), forming

population migration.

7 CONCLUSION

In this paper, principal component analysis is used to

rank the influence degree of many independent

variables which are linearly related to the dependent

variables. In view of the dimensionality reduction

idea of principal component analysis, this paper uses

the cluster method of independent variables. Through

the inter group variable correlation analysis, the

representative variables are taken from the unrelated

variable group for regression analysis, and the

dimensionality reduction effect similar to that of

principal component analysis can also be achieved.

The degree of explanation of independent variables to

dependent variables of the two methods is almost

equal.

ACKNOWLEDGEMENT

Supported by the young teacher development fund of

City College of Dongguan University of Technology

(2019QJY008Z);

Supported by Dongguan Science and Technology

of Social Development Program (20221800900952);

REFERENCES

Cai Fang. Future demographic dividend - the development

of sources of China's economic growth [J]. Chinese

Journal of Population Science, 2009, 2 (1): 4-12.

Chen Youhua. Demographic dividend and China's

economic growth [J]. Journal of Jiangsu Administration

Institute, 2008(4): 60-65.

Deng Weibin, Zhou Yumin et al. Practical course of

SPSS23 statistical analysis [M]. Beijing: Publishing

House of Electronics Industry, 2017.

Gao Jian, Wu Peilin. Impact of urban population size on

urban economic growth [J]. Urban Problems, 2016(6):

4-13.

Gao Guoli, Ji Renjun. Research on population migration in

the process of regional economic development - Taking

the Pearl River Delta region in Guangdong Province as

an example [J]. Economic Geography, 1995, 15(2): 76-

82.

Study on Inﬂuencing Factors of Population Changes in Dongguan City Based on Principal Component-Regression Analysis

195

He Xiaoqun. Multivariate statistical analysis [M]. Beijing:

China Renmin University Press, 2012.

Kong Weijun. On the influence of rural floating population

on urban-rural dual economic structure [J]. Guangxi

Social Sciences, 2001(1):129-132.

Li Wenxing, Zhang Zhengpeng. Forecast of Guangdong's

population trend under the "two-child" policy [J].

Journal of Guangzhou University (Social Science

Edition), 2018,17(10):89-97.

Liu Zhijia, Huang Heqing. Analysis of spatial-temporal

evolution characteristics of the interaction between the

expansion of construction land and economic and

demographic changes in the Pearl River Delta region

[J]. Resources Science, 2015, 37(7): 1394-1402.

Pan Wenxuan. Focus and path selection of the improvement

of structural tax reduction policy [J]. Taxation and

Economy, 2012(4): 67-71.

Smith. A. The wealth of Nations[M]. New York: Bantam

Classics,2003:35-40.

Wang Xiaoqin, Wang Hongmei." Demographic dividend"

effect and China's economic growth [J]. Economist,

2007(1): 104-110.

Wang You. Transformation of migration pattern of floating

population in Hebei Province and analysis of

influencing factors [D]. Hebei Normal University,

2020.

Wang Xiao'an. Cutting level determination of fuzzy graph

clustering [J]. Acta Botanica Boreali - Occidentalia

Sinica, 1998, 18(3): 445-449.

Wang Zhiyong. Population agglomeration and regional

economic growth - A test of Williamson hypothesis [J].

Social Sciences in Nanjing, 2018(3):60-69.

Yan Yueping, Huang Meixuan, Zheng Yiran. Research on

changes and trend of China’s population age structure

[J]. Dongyue Tribune, 2021, 42(1): 148-163.

Zhao Weihua. Law of urban population growth in China

and its enlightenment [J]. Journal of the Party School of

the Central Committee of the C.P.C, 2016, 20(3): 80-

85.

Zhong Shuiying, Li Kui. Summary of research on the

relationship between demographic dividend and

economic growth [J]. Population & Economics,

2009(2): 57-61.

Zhao Jing, Dan Qi. Mathematical modeling and

experiments. 4th edition [M]. Beijing: Higher

Education Press, 2014.

ICPDI 2022 - International Conference on Public Management, Digital Economy and Internet Technology

196