Frontiers in Business, Economics and Management 
ISSN: 2766-824X | Vol. 17, No. 3, 2024 

 

77 

GIS-based Analysis of Distribution Characteristics and 
Driving Factors of House Prices in Nanjing's Central 
District 

Xiaolin Gong * 

School of Public Administration, Central China Normal University, Wuhan, China 

* Corresponding author: Xiaolin Gong (Email: athenagxl@163.com) 

 

Abstract: This study aims to investigate the spatial distribution characteristics of house prices and their driving factors in the 
central area of Nanjing by using Geographic Information System (GIS) technology, combined with kernel density analysis, 

Moran index and geodetic detector method. The results show that: (1) The high values of house prices in central Nanjing have 

significant spatial aggregation characteristics, mainly concentrated in Qinhuai District, Gulou District and its border areas, and 

the house prices have significant spatial autocorrelation; (2) The results of the geo-detector show that the degree of traffic access 

and settlement characteristics have the greatest influence on house prices, while the influence of the construction of ancillary 

facilities is relatively weaker; (3) There are significant interactions between different factors, especially between transport 

accessibility and the construction of ancillary facilities, which play a crucial role in influencing house prices. 

Keywords: House price, Spatial differentiation, Geo-detector, Main city of Nanjing. 

 

1. Introduction 

As urbanization continues to accelerate, the issue of 

housing prices has become one of the hot issues of social 

concern. The issue of housing prices involves thousands of 

families, living with a home is one of the most basic people's 

livelihoods, the ‘CPC Central Committee on the formulation 

of the national economic and social development of the 14th 

Five-Year Plan and the 2035 Perspective Goals of the 

proposal’ clearly pointed out that we should promote the 

healthy development of housing consumption. Secondly, the 

stability and healthy development of real estate market prices 

is crucial to the national economy. Based on data published 

by the National Bureau of Statistics, residential real estate 

prices rose by 6.8% in 2023 compared to 2022, and although 

the rate of increase was reduced, the growth trend remained 

unchanged, and residential real estate prices in the first-tier 

cities rose by a large margin; the data also showed that real 

estate accounted for 5.8% of the entire China's GDP in 2023, 

and although this The data also suggests that real estate will 

account for 5.8% of China's GDP in 2023, and although this 

percentage has declined in recent years, it still has a 

substantial effect on the overall economy, suggesting that 

price volatility will not only affect the real estate sector, but 

also the financial markets and the economy as a whole. 

As an important first-tier city in eastern China, the trend of 

house prices in Nanjing's central city not only affects the 

quality of life of its citizens, but also reflects the overall level 

of urban development. This study is based on geographic 

information system (GIS) technology to conduct an in-depth 

analysis of the spatial distribution characteristics of housing 

prices and their driving factors in the central city of Nanjing, 

which is of great significance in the following aspects: firstly, 

it can help us to understand the differences in housing prices 

and their trends in different regions, analyse the causes of the 

differences, optimise the allocation of resources, and enhance 

the sense of well-being of residents' lives; secondly, it can 

explore in depth the various factors that affect housing prices 

and provide scientific basis for the government to formulate 

policies and developers to make business decisions; thirdly, 

through the study of the distribution of housing prices and its 

driving factors, it can help the government to better manage 

the real estate market, formulate reasonable land use and 

property regulation policies, and enhance the government's 

governance ability. 

Through reading the existing literature, it is found that most 

of the studies on real estate prices focus on socio-economic 

aspects, mostly discussing indicators such as population flow, 

land supply, economic development level, changes in 

government policies, and financial conditions [1-3]; most 

analyses of the current status of research on real estate prices 

start from the global, national, regional, or city-wide macro-

level [2,4-7], and the characteristics of spatial differentiation 

of house prices within cities are analysed quantitatively. 

quantitative analysis. In general, existing literature studies on 

specific range micro-regions in specific cities are not common, 

and the studies mainly reflect the pattern of spatial 

differentiation of house prices from an economic perspective, 

lacking the public participation attribute of the main body of 

house buying. 

In terms of research methodology, there have been few 

studies on real estate prices over the years, and their methods 

mostly use quantitative models to analyse the relationship 

between house prices and their influencing factors [8-10], 

mostly based on the Ordinary Least Squares (OLS) with the 

assumption of spatial mean, and the Geographically Weighted 

Regression (GWR) model based on spatial heterogeneity; 

however, the former does not take enough account of the 

spatial heterogeneity of house prices, and the latter takes 

account of the spatial heterogeneity but ignores the spatial 

influence scale difference of different variables, in this regard, 

some scholars also optimize the above problems and use 

MGWR model to conduct research [11]; in recent years, 

machine learning technology has been widely used in real 

estate research due to its unique advantage in processing 

massive data, and it has become a popular research method 



 

78 

[12-15]; based on this, many scholars at home and abroad 

have used machine learning methods to study factors such as 

traffic accessibility [16, 17], urban environment [18], and 

infrastructure support [19, 20], and to analyse their driving 

effects on housing prices. Analysing and summarising the 

above literature, it is found that geo-detector is less frequently 

used in the field of real estate price research. 

2. Data Sources and Research 
Methodology 

2.1. Data Sources 

The research data of this paper mainly includes 3 parts: 

First, based on the second-hand housing source information 

of Shell Housing (www.ke.com), a property information 

website in Nanjing, the total supply got 2974 housing source 

information, and other attribute data of the housing source 

were also obtained, including coordinates, streets, floors, etc.; 

second, through the Gaode open platform and the data 

crawling plug-in, the relevant point-of-interest data within the 

research scope was collected and filtered; Third, the 

population raster dataset with 100m resolution from China's 

seventh population census (2020) shared by Prof Chen 

Yuehong's team on the figshare platform. 

2.2. Research Methods 

2.2.1. Analysis of spatial distribution characteristics 
In this study, Kernel Density Analysis (KDA) is used to 

estimate the spatial density distribution of house prices in 

central Nanjing. By expanding each house price data point 

into a kernel function, a smooth density map is generated to 

show the spatial distribution of house prices. This approach 

can help identify hotspots in the distribution of second-hand 

houses for sale, assess the spatial distribution characteristics 

of house prices, as well as discover potential spatial patterns. 

The model equation is: 

𝑓(𝑥) =
1

𝑛ℎ
∑ 𝐾(

𝑥 − 𝑥𝑖
ℎ

)
𝑛

𝑖=1
 

Where f ̂(x) is the kernel density estimate at x; 

2.2.2. Spatial autocorrelation analysis 
Spatial autocorrelation analysis primarily involves global 

and local spatial autocorrelation, which is quantified using 

Moran's I. The model equations are as follows: 

𝐼 =
𝑛

𝑆0

∑ ∑ 𝑊𝑖𝑗 𝑍𝑖 𝑍𝑗
𝑛
𝑗=1

𝑛
𝑖=1

∑ 𝑍𝑖
2𝑛

𝑖=1

 

Where I am the Moran index, n is the total number of 

elements, Wij is the spatial weight between elements i and j, 

and S0 is the set of all spatial weights. 

2.2.3. Geographical Detector 
In this paper, we will refer to the Geographical Detector 

(GD) model of Wang Jinfeng [21] et al. to explore the spatial 

relationship and the mechanism of action between the 

distribution of house prices and its driving factors in the 

central city of Nanjing. The driving factors include 

accessibility, construction of supporting facilities and 

characteristics of the settlement. By dividing the study area 

into sub-areas, the geoprobes will measure the extent to which 

various factors explain the spatial distribution of house prices. 

The geodetector model includes a factor detector, an 

interaction detector, a risk detector, and an ecological detector: 

Factor detector: detects the impact of a single factor (e.g., 

housing conditions, traffic conditions, infrastructure support, 

etc.) on the spatial distribution of house prices, measured by 

the q-value, with the following model formula: 

𝑞 = 1 −
∑ 𝑁ℎ 𝜎ℎ

2𝐿
ℎ=1

𝑁𝜎 2
= 1 −

𝑆𝑆𝑊

𝑆𝑆𝑇
 

𝑆𝑆𝑊 = ∑ 𝑁ℎ

𝐿

ℎ=1

𝜎ℎ
2, 𝑆𝑆𝑇 = 𝑁𝜎 2 

Interaction detector: To detect the effect of the interaction 

between different factors on the house price distribution, the 

evaluation is done by firstly calculating the q-values of the 

two factors 𝑋1 and 𝑋2 on 𝑌: 𝑞(𝑋1) and 𝑞(𝑋2), respectively, 
and calculating the q-value of their interaction (the new 

polygonal distribution formed by the tangent of the two layers 

of the superimposed variables  𝑋1  and  𝑋2 ): 𝑞(𝑋1 ∩ 𝑋2) . 
Comparing 𝑞(𝑋1), 𝑞(𝑋2) with 𝑞(𝑋1 ∩ 𝑋2). 

Risk Detector: Identifies areas of high or low risk for higher 

or lower house prices and is tested using the t-statistic, 

modelled as follows: 

𝑡�̅�ℎ=1−�̅�ℎ=2 =
�̅�ℎ=1 − �̅�ℎ=2

[
𝑉𝑎𝑟(�̅�ℎ=1)

𝑛ℎ=1
+

𝑉𝑎𝑟(�̅�ℎ=2)
𝑛ℎ=2

]
1/2

 

Ecological detector: It is employed to determine whether 

there is a significant difference between the impacts of the 

two factors 𝑋1 and 𝑋2 on the spatial distribution of attribute 
𝑌, which is measured by the F-statistic, and the model formula 
is as follows: 

𝐹 =
𝑁𝑋1(𝑁𝑥2 − 1)𝑆𝑆𝑊𝑋1
𝑁𝑋2(𝑁𝑥1 − 1)𝑆𝑆𝑊𝑋2

 

𝑆𝑆𝑊𝑋1 = ∑ 𝑁ℎ 𝜎ℎ
2

𝐿1

ℎ=1

, 𝑆𝑆𝑊𝑋2 = ∑ 𝑁ℎ 𝜎ℎ
2

𝐿2

ℎ=1

 

Through the geo-detector model, the spatial distribution 

characteristics and their interrelationships affecting the house 

prices in the central area of Nanjing are analysed in depth, 

offering a scientific foundation for the formulation of targeted 

house price regulation and urban planning strategies. 

3. Characteristics of Spatial 
Differentiation of House Prices in 
The Main Urban Area of Nanjing 
City 

Firstly, the Arcgis kernel density analysis tool is used to 

derive the spatial differentiation characteristics of second-

hand houses for sale in the main urban area of Nanjing. 

According to the results of the analysis, it is found that the 

concentration areas of second-hand houses for sale in Nanjing 

are predominantly situated in the south-western part of the 

main city, namely Qinhuai District, Gulou District and their 

junction locations. 

 



 

79 

 
Figure 1. Kernel Density Analysis of House Prices in Nanjing's Central District 

In order to study the clustering characteristics of house 

prices within the main urban area of Nanjing, this study 

intends to use Moran scatter plot and LISA clustering map to 

start the analysis. The spatial autocorrelation is analysed by 

using GEODA global Moran index, and the resultant I-value 

is 0.4669, which reveals that the house prices in the main 

urban area of Nanjing city show the clustering characteristics. 

Subsequently, the local Moran index is used to analyse the 

local spatial autocorrelation type, and the clusters are first 

classified into four types: high - high, low - low, high - low, 

and low - high, and it is found that the high - high clusters of 

the house prices are mainly distributed in Gulou and Jianye 

districts as well as Xuanwu district, which are primarily 

located in the area bounded by Inner Ring South Road - 

Yingtian Avenue to the north, Inner Ring East Line to the west, 

and Beijing West Road to the south. This area features well-

known commercial districts such as Xinjiekou, with 

comprehensive supporting facilities, convenient 

transportation, and is also the administrative center of 

Nanjing. Most of the low-high agglomerations are distributed 

in a ring shape along the Xuanwu Lake Park, which has the 

unique advantage of a better ecological environment. The 

low-high agglomerations are often found in the surroundings 

of the high-high agglomerations, which may be affected by 

factors such as the ageing of the houses themselves and the 

surrounding congested living environment, leading to lower 

house prices. Low-low aggregations are mainly located in 

Yuhuatai District, the northern part of Xuanwu District, and 

Qinhuai District, which are generally less developed and are 

mostly surrounded by industrial and commercial land. 

 

 
Figure 2. LISA Cluster Map 

4. Factor Detection for Imaging 
Characteristics of House Price 
Divergence 

In order to explore the driving factors of spatial variation 

of house prices in Nanjing, this study uses geodetectors to 

calculate the influence factor data, and obtains the results of 

factor detection, interaction detection, and ecological 

detection. 

In this paper, the driving factors affecting the spatial pattern 

of house price divergence are classified into three categories, 

the first one is the degree of transport accessibility measured 

by the distance from metro stations, bus stops, and 

neighboring roads. The second category is the construction of 

ancillary facilities measured by the distance from 

neighbouring hospitals, educational institutions and shopping 

places. The third category is the characteristics of the 

neighbourhood measured by the population density of the 

street where it is located, the night-time lighting index (NLI) 

and the fractional vegetation cover (FVC). 

 

 

 

 



 

80 

Table 1. Impact factors 

 Drivers Factors 

X1 

Accessibility 

Subway station 

X2 Bus stops 

X3 Road 

X4 Construction of 

supporting 

facilities 

Shopping facilities 

X5 Medical institutions 

X6 Educational institutions 

X7 
Characteristics 

of the settlement 

Population density 

X8 Fractional vegetation cover (FVC) 

X9 Night light index (NLI) 

 
Based on the results of the factor probes, it is found that 

metro stations, bus stops, shopping places, hospitals, 

population, fractional vegetation cover, and nighttime 

lighting index are significant at the 5 per cent confidence level, 

with subway stations having a probability value (q) greater 

than 0.1 (0.113393), which has the greatest explanatory power 

for spatial divergence in house prices, followed by hospitals 

with a probability value close to 0.1 (0.092122), and shopping 

places and educational institutions have the weakest 

explanatory power for house prices. Overall, averaging the 

impact factors among the three categories of drivers reveals 

that the construction of neighbourhood amenities has the least 

impact on house price divergence, while the degree of 

transport access and settlement characteristics have a similar 

degree of impact on house price divergence, both being much 

higher than that of the impact brought by amenities. 

 

Table 2. The results of factor detector 

Factors p q Factors p q 

X1 0.113  0.000* X6 0.014  0.054  

X2 0.047  0.000* X7 0.063  0.000* 

X3 0.007  0.229  X8 0.032  0.000* 

X4 0.009  0.029* X9 0.058  0.000* 

X5 0.092  0.000*    

* Indicates significant at 5% confidence level 

 

Table 3. The results of ecological detector 

 X

1 
X2 X3 X4 X5 X6 X7 X8 

X

9 

X

1 
         

X

2 
N         

X

3 
Y N        

X

4 
Y N N       

X

5 
N N Y Y      

X

6 
Y N N N Y     

X

7 
N N N N N N    

X

8 
Y N N N N N N   

X

9 
N N N N N N N N  

 
Using an ecological detector to compare whether each 

factor has a significantly different impact on house prices, it 

is found that the difference in the impact of factors within 

each type of driver on house prices is not large, while there is 

a certain difference in the force of the factor between the 

degree of transport access and the construction of ancillary 

facilities, but overall, the impact of each factor on house 

prices is not significant. 

The interaction detector was utilized to identify the 

interactions among the drivers, and the results showed that the 

interaction force between any two of the nine selected 

influencing factors was greater than the influence of a single 

factor when it was in action. Among the drivers, X1∩X2, 

X4∩X5, X7∩X9 have the largest interaction power, 0.2185, 

0.1538, 0.1053 respectively; among the different types of 

drivers, X1∩X5, X1∩X7, X5∩aX7 have the largest 

interaction power, 0.2555, 0.1788, 0.1454 respectively. The 

maximum interaction force between accessibility and the 

construction of supporting facilities is greater than the 

maximum interaction force between the factors within itself. 

The factor interaction types include two-factor enhancement 

and non-linear enhancement, among which X1∩X9, X5∩X7, 

X5∩X9, X7∩X9, X8∩X9 show two-factor enhancement, and 

the rest show non-linear enhancement, which shows that the 

spatial pattern of house prices is the result of the joint action 

of different influencing factors. 

 
Figure 3. The results of interaction detector 

5. Conclusion 

This study analyses the spatial distribution pattern of house 

prices, as well as the global and local autocorrelation 

measures of house prices, using kernel density analysis and 

the Moran index, revealing the overall and local spatial 

correlation, aggregation and differentiation characteristics of 

house prices; and uses geo-detectors to identify spatial 

differentiation of house prices as well as the driving factors 

behind them. The main conclusions are as follows: 

Firstly, the house prices in central Nanjing present obvious 

spatial aggregation characteristics, with the main aggregation 

areas located in Qinhuai District, Gulou District and their 

junction positions in the southwestern part of the main city; 

the global Moran's I index is 0.4669, this indicates that house 

prices exhibit significant spatial autocorrelation, with high-

high clusters forming in Gulou District and Jianye District, 

and low-low clusters emerging in Yuhua District, northern 

Xuanwu District, and Qinhuai District. 

Secondly, the results of the factor probes show that 

transport accessibility and settlement characteristics have a 

more substantial effect on house prices, and in particular the 

proximity of metro stations and bus stops has the strongest 



 

81 

explanatory power for the spatial pattern of house price 

divergence. On the contrary, the impact of the construction of 

supporting facilities on house prices is relatively weak. 

Thirdly, the results of the interaction detector show that the 

interaction of different factors is obvious, and the interaction 

force between any two factors is greater than the influence of 

a single factor, especially there is a significant interaction 

between the degree of transport access and the construction 

of ancillary facilities; in addition, there is a complex 

relationship between the characteristics of the distribution of 

house prices and a variety of spatial factors, which indicates 

that the spatial distribution of house prices is the outcome of 

the combined influence of various factors.  

References 

[1] Li Kai, Fan Mingtai, Ye Sihui Research on the dynamic 
transmission of monetary factors and macro influence of 
housing prices in China —— Analysis based on TVP-SV-VAR 
model [J / OL]. Research on Financial Development, 2021 (1): 
29-37. DOI: 10.19647/j.cnki.37-1462/f.2021.01.004. 

[2] Zhang Long, Shen Yingqi China housing price stability 
measurement, influencing factor identification and monetary 
policy "cross-cycle" design [J]. Financial Science, 2023 (8): 1-
16. 

[3] Li Jiao, the heterogeneous fluctuation of urban housing prices 
in China: driving factors and action path [J]. Journal of 
Chongqing University of Technology (Social Sciences), 2023, 
37 (7): 93-109. 

[4] Yu Xiaofen, Zhan Dongsheng, Analysis of the spatial pattern 
and influencing factors of housing price index in 70 large and 
medium-sized cities in China [J / OL]. Journal of Central China 
Normal University (Humanities and Social Sciences Edition), 
2022,61 (1): 40-51. DOI: 10.19992/j.cnki.1000-
2456.2022.01.006. 

[5] Sun Biao, Yangshan, the characteristics and influencing factors 
of housing prices in big cities under the action of multi-centers 
—— Take Hefei city as an example [J]. Resources and 
Environment of the Yangtze River Basin, 2021, 30 (7): 1538-
1546. 

[6] Xu Danmeng, Li Xin, Zhang Suwen, Research on the spatial 
differentiation pattern and its influencing factors in Shenyang 
[J / OL]. Human Geography, 2021, 36 (6): 125-134. DOI: 
10.13959/j.issn.1003-2398.2021.06.014. 

[7] DUCA J V, MUELLBAUER J, MURPHY A. What drives 
house price cycles? International experience and policy issues 
[J]. Journal of Economic Literature, 2021, 59(3): 773-864. 

[8] Wu Chao, Liu Pengyu, Nie Ke The multi-scale spatial 
relationship between housing price and influencing factors in 
Nanjing [J]. Modern Urban Research, 2021 (4): 93-98. 

[9] Empirical evidence from 31 provinces and 70 large and 
medium-sized cities in China [J / OL]. Social Science in 
Nanjing, 2021 (6): 26-38. DOI: 10.15937/j.cnki.issn1001-
8263.2021.06.004. 

[10] Tu Jin, Jiang Wanchen, Leng Zhengxing The differentiation 
study of the influencing factors of the urban housing prices in 
China —— Based on the analysis of the big data of the second-
hand housing market in Chengdu city [J / OL]. Price Theory 
and Practice, 2021 (10): 75-78. DOI: 10.19851/j.cnki.cn11-
1010/f.2021.10.356. 

[11] Qiang Huanhuan, Wang Hui, Lei Shuyu Analysis on the spatial 
differentiation characteristics and influencing factors of 
housing prices in big cities based on MGWR model —— Take 
the main urban area of Nanjing as an example [J]. Modern 
Urban Research, 2024 (4): 109-116. 

[12] Chen Xiaoliang, Chen Standard, Wang Zhaorui, etc. The 
influencing factors of housing price differentiation among 
cities [J]. Economic Theory and Economic Management, 2024, 
44 (2): 49-64. 

[13] Chen Xiaoliang, Cheng Shuo, Chen Cao, and other studies on 
the influencing factors of housing prices in first-tier cities based 
on machine learning methods [J]. Nankai Journal (Philosophy 
and Social Sciences edition), 2023 (6): 146-163. 

[14] MAJ, CHENG J C, JIANG F, et al. Analyzing driving factors 
of land values in urban scale based on big data and non-linear 
machine learning techniques [J]. Land use policy, 2020, 94: 
104537. 

[15] HO W K O, TANG B S, WONG S W. Predicting property 
prices with machine learning algorithms [J/OL]. Journal of 
Property Research, 2021, 38(1): 48-70. 
DOI:10.1080/09599916.2020.1832558. 

[16] YANG L, CHAU K W, SZETO W Y, et al. Accessibility to 
transit, by transit, and property prices: Spatially varying 
relationships [J]. Transportation Research Part D: Transport 
and Environment, 2020, 85: 102387. 

[17] YANG L, CHUX, GOUZ, et al. Accessibility and proximity 
effects of bus rapid transit on housing prices: Heterogeneity 
across price quantiles and space [J]. Journal of Transport 
Geography, 2020, 88: 102850. 

[18] BISHOP K C, KUMINOFF N V, BANZHAF H S, et al. Best 
Practices for Using Hedonic Property Value Models to 
Measure Willingness to Pay for Environmental Quality [J/OL]. 
Review of Environmental Economics and Policy, 2020, 14(2): 
260-281. DOI:10.1093/reep/reaa001. 

[19] GUIFFARD J B. Valuing the virtual: The impact of fiber to the 
home on property prices in France [J]. Telecommunications 
Policy, 2024, 48(4): 102732. 

[20] Fu Yangjun, Qu Weidong advance and retreat: Research on 
urban Residence Decision of Floating Population —— 
Analysis the effect of multiple factors such as urban housing 
price and public service level [J / OL]. Price Theory and 
Practice, 2023 (7): 131-135. DOI: 10.19851/j.cnki.cn11-
1010/f.2023.07.187. 

[21] Wang Jinfeng, Xu Chengdong: Principle and Prospect [J]. 
Journal of Geography, 2017,72 (1): 116-134.