# RECORD 84fe25f12d7833b71a83f2c2ae3a06dfc013ffcf
# TITLE: Do street-level scene perceptions affect housing prices in Chinese megacities? An analysis using open access datasets and deep learning
# YEAR: 2019.0  VENUE: PLoS ONE  DOI: 10.1371/journal.pone.0217505
# AUTHORS: Xiao Fu; Tian-Xia Jia; Xue-Qi Zhang; Shanlin Li; Yonglin Zhang
# PAGES: 18  CANDIDATE TABLE PAGES (1-based): [9, 10, 13]
# ABSTRACT: Many studies have explored the relationship between housing prices and environmental characteristics using the hedonic price model (HPM). However, few studies have deeply examined the impact of scene perception near residential units on housing prices. This article used house purchasing records from FANG.com and open access geolocation data (including massive street view pictures, point of interest (POI) data and road network data) and proposed a framework named “open-access-dataset-based hedonic price modeling (OADB-HPM)” for comprehensive analysis in Beijing and Shanghai, China. A state-of-the-art deep learning framework and massive Baidu street view panoramas were employed to visualize and quantify three major scene perception characteristics (greenery, sky and building view indexes, abbreviated GVI, SVI and BVI, respectively) at the street level. Then, the newly introduced scene perception characteristics were combined with other traditional characteristics in the HPM to calculate marginal prices, and the results for Beijing and Shanghai were explored and compared. The empirical results showed that the greenery and sky perceptual elements at the property level can significantly increase the housing price in Beijing (RMB 39,377 and 6011, respectively) and Shanghai (RMB 21,689 and 2763, respectively), indicating an objectively higher willingness by buyers to pay for houses that provide the ability to perceive natural elements in the surrounding environment. This study developed quantification tools to help decision makers and planners understand and analyze the interaction between residents and urban scene components.


===== PAGE 1 (score 5, numeric-paren 0, table-words 0, var-words 5) =====

RESEARCH ARTICLE
Do street-level scene perceptions affect
housing prices in Chinese megacities? An
analysis using open access datasets and deep
learning
Xiao Fu1, Tianxia Jia1,2, Xueqi Zhang1,2, Shanlin Li3, Yonglin ZhangID1,2*
1 State Key Laboratory of Urban and Regional Ecology, Research Center for Eco-Environmental Sciences,
Chinese Academy of Sciences, Beijing, China, 2 College of Resources and Environment, University of
Chinese Academy of Sciences, Beijing, China, 3 National Science Library, Chinese Academy of Sciences,
Beijing, China
* ylzhang_rcees@foxmail.com
Abstract
Many studies have explored the relationship between housing prices and environmental
characteristics using the hedonic price model (HPM). However, few studies have deeply
examined the impact of scene perception near residential units on housing prices. This arti-
cle used house purchasing records from FANG.com and open access geolocation data
(including massive street view pictures, point of interest (POI) data and road network data)
and proposed a framework named “open-access-dataset-based hedonic price modeling
(OADB-HPM)” for comprehensive analysis in Beijing and Shanghai, China. A state-of-the-
art deep learning framework and massive Baidu street view panoramas were employed to
visualize and quantify three major scene perception characteristics (greenery, sky and build-
ing view indexes, abbreviated GVI, SVI and BVI, respectively) at the street level. Then, the
newly introduced scene perception characteristics were combined with other traditional
characteristics in the HPM to calculate marginal prices, and the results for Beijing and
Shanghai were explored and compared. The empirical results showed that the greenery
and sky perceptual elements at the property level can significantly increase the housing
price in Beijing (RMB 39,377 and 6011, respectively) and Shanghai (RMB 21,689 and 2763,
respectively), indicating an objectively higher willingness by buyers to pay for houses that
provide the ability to perceive natural elements in the surrounding environment. This study
developed quantification tools to help decision makers and planners understand and ana-
lyze the interaction between residents and urban scene components.
Introduction
Urban scenes can be considered a vital medium for human beings, and it is important to
recognize and understand this aspect of cities [1] to ensure the spatial security of urban
sustainability [2]. The spatial distributions of urban scenes exhibit variety, complexity and
PLOS ONE | https://doi.org/10.1371/journal.pone.0217505
May 30, 2019
1 / 18
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Fu X, Jia T, Zhang X, Li S, Zhang Y (2019)
Do street-level scene perceptions affect housing
prices in Chinese megacities? An analysis using
open access datasets and deep learning. PLoS
ONE 14(5): e0217505. https://doi.org/10.1371/
journal.pone.0217505
Editor: Shihe Fu, Xiamen University, CHINA
Received: January 16, 2019
Accepted: May 12, 2019
Published: May 30, 2019
Copyright: © 2019 Fu et al. This is an open access
article distributed under the terms of the Creative
Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in
any medium, provided the original author and
source are credited.
Data Availability Statement: All relevant data are
within the manuscript and its Supporting
Information files.
Funding: All authors were funded by The National
Key Research and Development Program of China
(2016YFC0503605). The funders had no role in
study design, data collection and analysis, decision
to publish, or preparation of the manuscript.
Competing interests: The authors have declared
that no competing interests exist.


===== PAGE 9 (score 31, numeric-paren 0, table-words 3, var-words 15) =====

descriptions of the variables are shown in Table 1. The data frames used to build the HPMs
can be found at https://doi.org/10.6084/m9.figshare.7823975.v1.
The independent variables in this article were classified into four major categories: location,
structure, neighborhood and scene perception characteristics. The location characteristic
included one variable named CENTER, which represented the shortest road distance from a
residential unit to the city center. To be more realistic, road distance and road network buffers
were adopted for all distance calculations in this paper.
The structure characteristics included AREA (average usable area in the apartment), year
(year of construction), HS (households), FR (floor-area ratio), PF (property management fee),
GR (greening rate) and the dummy variables ORI and BU. The dummy variable ORI repre-
sented the main orientation of the buildings in the residential unit: if the buildings mainly face
south, the value was set to 1; otherwise, it was set to 0. The dummy variable BU represented
the major building type: if the building was a slab, the value was set to 1; otherwise, it was set
to 0.
Table 1. Variable descriptions and basic statistical information.
Cities
Beijing
Shanghai
Variables
Description
Mean Standard deviation Mean Standard deviation
Dependent variable
LNPRICE
Log selling price in 10,000 RMB (Chinese currency, US $1 = RMB 6.71)
5.64
0.54
5.43
0.58
Location characteristic
CENTER
Road distance to city center (km)
10.87
6.32
15.11
9.30
Structure characteristics
AREA
Average usable area in the apartment (m2)
88.45
96.84
86.58
44.97
AGE
2018 minus the year of construction of the building
19.64
13.75
22.78
23.90
ORI
Dummy variable, 1 if the building has windows facing south
0.78
0.41
0.93
0.25
HS
Number of households
1030
1200
1037
942
FR
Floor-area ratio
2.52
1.55
2.27
1.61
PF
Property management fee (RMB/m2 per month)
1.76
1.42
1.27
1.28
GR
Green coverage rate (%)
0.32
0.07
0.34
0.11
BU
Dummy variable, 1 if the building is a slab; 0 if the building is any other
type
0.60
0.49
0.87
0.34
Neighborhood characteristics
AIRP
Road distance to the nearest airport (km)
22.13
6.75
13.56
8.58
BUS
Road distance to the nearest bus station (km)
0.24
0.20
0.18
0.16
SUB
Road distance to the nearest subway station entrance (km)
1.47
1.33
1.63
2.11
TRAIN
Road distance to the nearest train station (km)
4.74
3.04
7.09
5.26
FINAN
Road distance to the nearest financial facility (km)
0.26
0.36
0.32
0.36
RESTA
Road distance to the nearest restaurant (km)
0.14
0.21
0.15
0.22
HOSP
Road distance to the nearest hospital (km)
0.18
0.22
0.23
0.24
EDU
Road distance to the nearest educational facility (km)
0.18
0.24
0.21
0.24
SHOP
Road distance to the nearest shopping mall (km)
0.10
0.17
0.09
0.16
PARK
Road distance to the nearest park (km)
1.66
1.34
1.46
1.08
WATER
Road distance to the nearest water body (km)
0.73
0.50
0.35
0.28
WATER_A
Area of the nearest river or lake (km2)
0.12
0.22
0.11
0.28
Scene perception
characteristics
LNGVI
Logarithmic mean green view index within 800 m road distance
2.80
0.34
2.97
0.31
LNBVI
Logarithmic mean building view index within 800 m road distance
2.22
0.54
2.37
0.55
LNSVI
Logarithmic mean sky view index within 800 m road distance
3.74
0.16
3.65
0.26
https://doi.org/10.1371/journal.pone.0217505.t001
Street-level scene perceptions affect housing prices in typical Chinese megacities
PLOS ONE | https://doi.org/10.1371/journal.pone.0217505
May 30, 2019
9 / 18


===== PAGE 10 (score 26, numeric-paren 0, table-words 4, var-words 15) =====

The neighborhood characteristics included the road distances between the housing estates
and diverse urban amenities and environmental infrastructures, including the shortest road
distances to airports (AIRP), bus stations (BUS), subway stations (SUB), train stations
(TRAIN), financial facilities (FINAN), restaurants (RESTA), medical facilities (HOSP), educa-
tional facilities (EDU), commercial facilities (SHOP), urban parks (PARK), and water bodies
(WATER), as well as the area of the nearest water body (WATER_A).
Finally, the newly introduced scene perception characteristics (GVI, BVI and SVI) at the
residential unit level were transformed by the natural logarithm before building the HPMs.
The coefficients of the independent variables can explain the percentage change in the depen-
dent variables in the HPMs.
Results and discussion
Mapping scene perception indicators
To improve the observations and comparisons, this study plotted the street-level GVI, BVI and
SVI values in thematic maps for Beijing and Shanghai in 2015, as shown in Fig 5. The three view
indexes (VIs) were each mapped using five value intervals, and the value intervals in Beijing and
Shanghai were the same. The GVI value intervals were [0.00, 0.10], [0.11, 0.20], [0.21, 0.30], [0.31,
0.40], and [0.41, 0.70]; the BVI intervals were [0.00, 0.05], [0.06, 0.10], [0.11, 0.20], [0.21, 0.30], and
[0.31, 0.65]; and the SVI intervals were [0.00, 0.20], [0.21, 0.35], [0.36, 0.45], [0.46, 0.55], and [0.56,
0.65], which correspond to the qualitative levels of very low, low, medium, high and very high.
The statistical results indicate that the average GVI values in Beijing and Shanghai are 16.44
and 19.49, respectively, which indicated that pedestrians in Shanghai tend to see more green
scenes than those in Beijing. Fig 5(A) and 5(B) indicate that high and very high GVI road seg-
ments have a broader distribution and in general look “greener” in Shanghai than in Beijing.
The GVI values of road segments above the high level are mainly located in the central areas
(Dongcheng and Xicheng Districts), the northwestern area (Haidian District) and part of the
southern area, but the high GVI segments are mainly concentrated along the western bank of
the Huangpu River in Shanghai. The statistical results indicate that the average BVI value in
Shanghai is slightly higher than that in Beijing (10.7 and 9.21, respectively). The spatial distri-
butions in Fig 5(C) and 5(D) show that the BVI values in both cities present low-inside-and-
high-outside gradient variations. The spatial distributions of the SVI values in Beijing and
Shanghai show the opposite pattern relative to the BVI values (Fig 5(E) and 5(F)) and present
high-inside-and-low-outside gradient variations. According to the statistical results, the aver-
age SVI of Beijing is higher than that of Shanghai (42.10 and 38.47, respectively). Therefore,
from an overall perspective, people in Shanghai tend to see more profile areas of greenery and
buildings, whereas people in Beijing tend to see more open road canyons.
OLS regression modeling
The OLS regression modeling was four-fold: first, all independent variables were incorporated
into the OLS model calculation; second, all VIs were removed from the model, but other vari-
ables remained the same; third, the GVI was added to the OLS model; fourth, the GVI and
BVI were added to the model. The last three models controlled the VIs to show their contribu-
tions to the adjusted R2, and we included the last three models in the Supporting Information
to show the soundness of our model (S1 File).
The model results in Table 2 show that the adjusted R2 values were 0.7242 and 0.7511 for Bei-
jing and Shanghai, respectively, and most of the independent variables were statistically
significant. The variance inflation factor (VIF) test was performed to verify whether the variables
exhibited multicollinearity. The VIFs of the variables (less than 4) indicated that the
Street-level scene perceptions affect housing prices in typical Chinese megacities
PLOS ONE | https://doi.org/10.1371/journal.pone.0217505
May 30, 2019
10 / 18


===== PAGE 11 (score 0, numeric-paren 0, table-words 0, var-words 0) =====

Fig 5. The spatial distributions of VIs in Beijing and Shanghai (GVI, BVI and SVI).
https://doi.org/10.1371/journal.pone.0217505.g005
Street-level scene perceptions affect housing prices in typical Chinese megacities
PLOS ONE | https://doi.org/10.1371/journal.pone.0217505
May 30, 2019
11 / 18


===== PAGE 13 (score 40, numeric-paren 0, table-words 4, var-words 16) =====

housing prices through the newly built OADB-HPM framework. The model results in the pre-
vious section showed that street-level GVI and SVI had positive effects on the neighboring
housing prices in both Beijing and Shanghai, indicating that house buyers may have require-
ments for urban greenery and sky openness; however, BVI was not significant in any model,
indicating that house buyers did not favor more visible artificial buildings in their living envi-
ronment. According to the marginal price calculation, an increase in GVI and SVI by 1% can
increase housing prices by 0.2273% and 0.0899%, respectively, in Beijing and by 0.1853% and
0.0466%, respectively, in Shanghai (the conversion formula for the marginal prices for the
Table 2. OLS regression model (all variables).
Model 1 (Beijing)
Model 2 (Shanghai)
Variables
Full name
Unstandardized
Coefficients
Standard Error
VIF
Unstandardized
Coefficients
Standard Error
VIF
Constant
Constant
3.828
0.4116
5.55
0.2933
Location characteristic
CENTER
City center
-0.0452
0.0012
2.51
-0.0154
0.0012
3.45
Structure characteristics
AREA
Housing area
0.0079
0.0002
1.55
0.0099
0.0001
1.39
AGE
Building age
-0.0027
0.0001
1.19
-0.0001
0.0001
1.05
ORI
Orientation
0.0679
0.0131
1.13
0.0532
0.0204
1.02
HS
Households
0.0000
0.0000
1.36
0.0000
0.0000
1.49
FR
Floor-area ratio
-0.0084
0.0034
1.18
0.0000
0.0000
1.01
PF
Property management fee
0.0324
0.0054
1.47
0.0021
0.0028
1.13
GR
Green coverage rate
0.2421
0.0822
1.18
0.2744
0.0832
1.35
BU
Building types
0.0322
0.0119
1.21
0.0189
0.0225
1.05
Neighborhood characteristic
AIRP
Airport
0.0176
0.0091
1.17
-0.0077
0.0013
3.46
BUS
Bus station
-0.0117
0.0315
1.15
-0.0447
0.0392
1.14
SUB
Subway station
-0.0172
0.0048
1.59
-0.0073
0.0039
1.60
TRAIN
Train station
0.0024
0.0022
1.83
-0.0006
0.0013
1.88
FINAN
Financial facility
-0.0858
0.0189
1.94
-0.0892
0.0190
1.78
RESTA
Restaurant
0.1017
0.0389
2.18
0.0421
0.0511
1.65
HOSP
Hospital
0.0151
0.0360
1.90
0.0116
0.04134
1.85
EDU
Educational facility
-0.0614
0.0330
1.46
-0.0958
0.0362
1.69
SHOP
Shopping mall
0.1149
0.05178
2.69
0.1042
0.0677
2.70
PARK
Park
-0.0033
0.0048
1.62
-0.0153
0.0059
1.40
WATER
Water body
0.0337
0.0102
1.16
0.0416
0.0246
1.22
WATER_A
Water body area
0.0447
0.0215
1.12
0.0226
0.0367
1.08
Scene perception characteristics
LNGVI
Green view index
0.2273
0.0818
3.31
0.1853
0.0471
2.71
LNBVI
Building view index
-0.0005
0.0018
3.96
-0.0008
0.0014
2.64
LNSVI
Sky view index
0.0899
0.0280
3.17
0.0466
0.0298
3.50
F ratio
200.5200
171.1500
Adjusted R2
0.7242
0.7511
Durbin-Watson
1.7628
1.8212
 Indicates significance at the 10% level
 indicates significance at the 5% level
 indicates significance at the 1% level.
https://doi.org/10.1371/journal.pone.0217505.t002
Street-level scene perceptions affect housing prices in typical Chinese megacities
PLOS ONE | https://doi.org/10.1371/journal.pone.0217505
May 30, 2019
13 / 18


===== PAGE 14 (score 17, numeric-paren 0, table-words 1, var-words 15) =====

dummy logarithm transformed variables were ex−1, where x is the nonnormalized coefficient
in the model). Therefore, the house buyers in Beijing and Shanghai favor the perception of nat-
ural aspects (green vegetation and open sky).
Through a comparison of the model results, home buyers in Beijing and Shanghai exhibited
the same willingness to pay for the GVI and SVI, but there were significant differences in the
marginal prices. Residents in Beijing showed significantly more willing to pay for GVI than
those in Shanghai, which may be caused by the locations, climates and seasons of the two cities.
Moreover, the two cities were located at different latitudes, and the green leaf period in Beijing
is shorter than that in Shanghai due to climate, humidity and seasonal changes. This difference
may be one of the major reasons affecting people’s perception and willingness to pay for the
greenery around housing estates.
In summary, according to the results in this article, the government should attach impor-
tance to the adjustment and optimization of the visual scenery around residential areas and
high-rise building clusters to improve people’s visual access to natural elements. For example,
the construction of open spaces and green parks in the public spaces between neighboring
housing estates may help to meet the needs of people. The government can fine-tune and
reshape the urban spatial structure according to the needs of residents to build an appropriate
living environment to meet people’s practical needs. For instance, the configuration of three-
dimensional green walls, corridors and roads can increase the daily interaction opportunities
between people and green vegetation, and the appropriate use of domes made of transparent
material and open walls may increase the sense of openness in various spaces.
The government should pay attention to the potential economic impacts of urban scene
perception on real estate. On the one hand, in the absence of government control, real estate
developers could exploit urban green and open spaces for additional benefits. Over time, the
excessive accumulation of such benefits may lead to low-income consumers being unable to
bear the financial costs resulting from increased housing prices. This phenomenon is detri-
mental to the sustainable development of the environment, society and economy. This phe-
nomenon not only violates the fairness of the living environment in cities but also may harm
the overall interests of citizens. On the other hand, in terms of the pleasant visual environment
surrounding a property, developers almost unconditionally obtain the economic benefits of a
pleasant residential environment but do not contribute to the maintenance or optimization of
the urban visual landscape. Therefore, the government should collect real estate tax from real
estate developers according to the constructions of urban scenes, and this revenue could be
used to fund the optimization and maintenance of urban scenes. The government needs to
strengthen the supervision of housing sales in the real estate market, restrain behaviors that
abuse green scenes and open spaces as bargaining chips in the process of real estate sales to dis-
rupt the market, and prevent developers from maliciously raising housing prices by citing the
appropriateness of the public visual environment during the sales process. The goals of real
estate developers should be to properly use the influence of urban visual elements on housing
price on the basis of reasonable bidding and scientific planning, improve the competitiveness
of their real estate brands, and participate in the construction of high-quality residential
environment.
In summary, this article identified the actual requirements for visible greenery and sky
openness in Beijing and Shanghai based on the newly constructed OADB-HPM framework.
Due to the previous lack of technology and appropriate research framework, the potential
effect of the visible element characteristics of residential environments on housing prices has
often been ignored by researchers and decision makers. However, the framework and techni-
cal means proposed in this paper can now be used as an effective supplement and quantifica-
tion tool.
Street-level scene perceptions affect housing prices in typical Chinese megacities
PLOS ONE | https://doi.org/10.1371/journal.pone.0217505
May 30, 2019
14 / 18