1# RECORD cfc27a33426e6615c17a451873b0b54c656122522# TITLE: Multilevel effects of urban form and urban functional zones on housing prices: evidence from open-source big data3# YEAR: 2024.0 VENUE: Journal of Housing and the Built Environment DOI: 10.1007/s10901-023-10109-y4# AUTHORS: Guan Huang; Si Qiao; A. Yeh5# PAGES: 25 CANDIDATE TABLE PAGES (1-based): [11, 12, 16, 17, 21]6# ABSTRACT: While the relationship between local housing prices and the urban form and distribution of urban functional zones in a single city is well-discussed, the conclusion is usually sensitive to a particular city context, and cross-city comparative study is limited. This study attempts to examine the influences of urban form and urban functional zone distribution on housing prices within and between cities after controlling the city-wide socio-economic and demographic differences. Based on multiple open-source big data, such as points-of-interest (POI) and historical housing transaction data, the hierarchical linear model is utilized to compare the housing market of 10 extra-large cities in China. Results indicate that the urban form and the urban functional zone distribution significantly influence housing prices after the socio-economic and demographic differences are controlled. For inter-city comparison, an urban form with high compactness, low centrality, low polycentricity, high density, and low dissimilarity in housing development is related to lower city-level housing prices. For intra-city, proximity to work centers, high-quality hospitals, and schools shows positive associations to housing prices.789===== PAGE 1 (score 11, numeric-paren 0, table-words 0, var-words 11) =====1011Vol.:(0123456789)12Journal of Housing and the Built Environment (2024) 39:987–101113https://doi.org/10.1007/s10901-023-10109-y141 315ARTICLE16Multilevel effects of urban form and urban functional zones 17on housing prices: evidence from open‑source big data18Guan Huang1 · Si Qiao1 · Anthony Gar‑On Yeh119Received: 1 November 2022 / Accepted: 22 December 2023 / Published online: 14 February 2024 20© The Author(s) 202421Abstract22While the relationship between local housing prices and the urban form and distribution of urban 23functional zones in a single city is well-discussed, the conclusion is usually sensitive to a par-24ticular city context, and cross-city comparative study is limited. This study attempts to examine 25the influences of urban form and urban functional zone distribution on housing prices within 26and between cities after controlling the city-wide socio-economic and demographic differences. 27Based on multiple open-source big data, such as points-of-interest (POI) and historical housing 28transaction data, the hierarchical linear model is utilized to compare the housing market of 10 29extra-large cities in China. Results indicate that the urban form and the urban functional zone 30distribution significantly influence housing prices after the socio-economic and demographic 31differences are controlled. For inter-city comparison, an urban form with high compactness, 32low centrality, low polycentricity, high density, and low dissimilarity in housing development is 33related to lower city-level housing prices. For intra-city, proximity to work centers, high-quality 34hospitals, and schools shows positive associations to housing prices.35Keywords Urban form · Urban functional zone · Housing prices · Natural semantical 36language process · Multilevel effect · Open-source big data371 Introduction38Research on housing prices based on the hedonic model has widely clarified that the 39properties of houses (square footage, number of bedrooms, house orientation, building 40age, and so on), the location, and the neighborhood characteristics undoubtedly affect 41the prices (Goodman, 1978; Rosen, 1974). Typically, these empirical studies state them-42selves as unique to the experience of a given city and attribute the uniqueness to the 43 * Si Qiao 44 45siqiao@connect.hku.hk46 47Guan Huang 48 49guanhuang@connect.hku.hk50 51Anthony Gar‑On Yeh 52 53hdxugoy@hku.hk541 55Department of Urban Planning and Design, The University of Hong Kong, Pokfulam Road, 56Hong Kong SAR, China575859===== PAGE 11 (score 9, numeric-paren 0, table-words 2, var-words 3) =====606199762Multilevel effects of urban form and urban functional zones…631 3643.3 Potential determinants65Several potential influencing factors possibly affect the HEPrice even after internalizing 66the housing structure attributes’ effects. According to previous literature reviews, we clas-67sified three categories of indicators to evaluate the cross-scale potential determinants of 68housing estate price (see Table 3).69Category 1 is the individual category, including the structure and location attributes of 70specific housing estates. Estate building year has a significant impact on housing price due 71to the bank loan policy in China. A high down payment is associated with an older house, 72a shorter load period, and a lower available loan amount. The increasing burden of down 73payment with the increased building year would intensely affect buyers’ choice and thus be 74reflected in the pricing strategy of sellers. Furthermore, the location attributes have a wide 75impact on the capability of accessing urban service opportunities and access to urban ser-76vices may increase house values, as evidenced by higher prices for homes close to transit in 77previous studies (Duncan, 2011; Shen & Karimi, 2017).78Table 2 Descriptive statistics of 79average per-square-meter housing 80estate price in the ten extra-large 81cities82# : Price unit: yuan/m2, CNY. The currency exchange rate is 100 CNY 83to 15.65 USD for 22nd Oct 202184City name85Min86Max87Mean88Std. Dev89Counts90Shanghai9110,62992170,319936,5061.699431,560.97959,45496Beijing9717,93798172,414996,5477.4910034,375.781015,966102Shenzhen10324,166104131,0001056,1640.4210623,162.431073,432108Guangzhou1097,87911093,7591113,5901.1511220,316.751134,396114Chengdu1155,08911630,03211712,035.4911845,65.7321198,026120Hangzhou1219,915122100,79912339,198.3812419,098.931253,310126Wuhan1275,91812832,65412916,397.6813055,36.6131314,098132Nanjing1339,41813483,83613534,011.2713615,025.991373,749138Changsha1395,35614024,58214110,011.9814239,25.4671432,019144Xi’an1456,09314640,08414714,086.0114861,03.1781492,956150Fig. 4 Interpolated housing prices. (a) Beijing, (b) Xi’an, (c) Chengdu, (d) Shanghai, (e) Nanjing, (f) 151Hangzhou, (g) Wuhan, (h) Changsha, (i) Guangzhou, (j) Shenzhen152153154===== PAGE 12 (score 14, numeric-paren 0, table-words 2, var-words 8) =====155156998157 158G. Huang et al.1591 3160Table 3 Variables and statistics based on three categories of spatial scales161Category162Sub-category163Variables [unit]164Mean165Std. dev166Level1671. Housing estate indicators168Estate structure attribute169Building year [year]1701999.6717111.487172Level 1173Estate location attribute174Proximity to metro station [kilometer]1751.571762.008177Level 1178Proximity to bus station [kilometer]1790.271800.136181Level 11822. Intra-city indicators183Estate-based urban functions attribute184Proximity to working zone [kilometer]1850.531860.386187Level 1188Proximity to shopping center [kilometer]1891.141900.715191Level 1192Proximity to commercial business zone [kilometer]1932.741942.276195Level 1196Proximity to park and green space [kilometer]1971.571981.788199Level 1200Proximity to 3A hospital [kilometer]2015.002025.462203Level 1204Proximity to primary and middle school [kilometer]2050.652060.426207Level 12083. Inter-city indicators209City-based urban form attribute210Compactness [index]2110.13212.06213Level 2214Centrality [index]2150.71216.116217Level 2218Polycentricity [index]2193.972201.683221Level 2222Building area ratio [ratio]2231.47224.242225Level 2226Dissimilarity [index]2270.432280.08229Level 2230City-based socio-economic attribute231Gini coefficient of population [continuous]2320.51233.085234Level 2235Moran index of education [continuous]2360.20237.076238Level 2239GDP per capita [1 million Yuan/population]2400.18241.022242Level 2243244245===== PAGE 13 (score 28, numeric-paren 0, table-words 7, var-words 14) =====246247999248Multilevel effects of urban form and urban functional zones…2491 3250Category 2 attempts to explore the effects of the distribution of urban functional zones 251in intra-city, which is an extension of the impact of housing estates on the neighborhood 252attributes. The proximity to urban functional zones represents the time cost and expense 253burden of reaching potential destinations to obtain urban services, such as medical service, 254educational resources, and physical exercise amenities. The proximity indicators of urban 255functional zones, including working zones, shopping centers, commercial business zones, 256parks and green spaces, 3A hospitals, and primary and middle schools, are selected as the 257level 1 indicators (Qiao et al., 2021; Tam et al., 2019; Wang et al., 2017; Wen et al., 2017; 258Zhang et al., 2023).259Category 3 covers the city-level attributes of the socio-economic and physical urban 260form. The physical urban form indicators include the compactness, centrality, polycentric-261ity, dissimilarity, and density. The details are presented in Sect. 2. The socio-economic fac-262tors include the Gini coefficient of population, the Moran index of education, and the GDP 263per capita, (Tsai, 2005; Villar & Raya, 2015; Zhang et al., 2016). The Gini coefficient of 264population represents the unequal population distribution in the metropolitan area. A high 265Gini coefficient approaching 1 indicates a high unevenness of population. The Moran index 266of education ranges from -1 to 1, illustrating the clustering pattern for the low-educated 267population (educated in primary school or below). If the value approaches 0, then the pat-268tern is randomly scattered, while an absolute value near 1 represents a clustered pattern. 269This indicator shows whether low education, as a proxy of poverty, is agglomerated. The 270GDP per capita indicates a city’s level of economic development. As the urban form indi-271cator of building density is collinear with the population density, we omitted it. The final 272variables and statistics are shown in Table 3. The variance inflation factor (VIF) of the 273final variables is provided in Appendix Table A3. The VIFs of majority of variables are 274less than 5 and all VIFs are less than 10 indicating there is no significant multicollinearity 275between the final variables.2763.4 Hierarchical regression model277The traditional ordinary least squares (OLS) hedonic model has been criticized for ignor-278ing the existence of housing market segmentation (Goodman & Thibodeau, 1998, 2003). 279In short, a house occupies a unique location in space, and the geographic attributes tied to 280its location cannot be replicated. The demand for houses in certain locations may be very 281inelastic, such as houses near prestigious schools (Wen et al., 2017). The irreplaceability 282of different housing locations shows that houses are essentially nested in spatial submar-283kets with different pricing premiums (Orford, 2000). Therefore, to solve the fixed effects 284of intra-city and inter-city submarkets, the hierarchical linear model (HLM) is introduced. 285In the first level, the individual housing estate attributes of proximity to urban functional 286zones are measured (see Table 3, “Level” column). This level reflects the housing values 287nested in intra-city functions. In the second level, the inter-city attributes of the urban phys-288ical forms and the city-based socio-economics of the habitants are assessed. The formulas 289of the two-level HLM used in this study are as follows (Goodman & Thibodeau, 2003):290where Pij is the logarithm-transformed HEPrice of estate i within city j; 훼pj (p = 0, 1, 2, …, 291p) is an intercept and regression coefficient. x1ij is the level 1 variable that represent indi-292vidual housing estate attributes. eij is a random error.293Level1 ∶Pij = 훼0j + 훼1j294(x1ij295) + 훼2j296(x2ij297) + ⋯+ 훼pj298(xpij299) + eij, eij ∈N(0, 휎2),300301302===== PAGE 16 (score 65, numeric-paren 34, table-words 5, var-words 10) =====3033041002305 306G. Huang et al.3071 3308Table 5 Effects of urban forms and urban functional zones on housing estate prices at multi levels309*** p < .01, **p < .05, *p < .1310Price (log)311Model 1312Model 2313Model 3314OLS model315Coef316Robust Std.Err317Coef318Robust Std.Err319Coef320Robust Std.Err321Coef322Robust Std.Err323Building year [level 1]324 − 0.007***3250.002326 − 0.0013270.003328 − 0.0013290.0033300.000**3310.000332Proximity to metro station [level 1]333 − 0.0083340.007335 − 0.0083360.007337 − 0.010***3380.002339Proximity to bus station [level 1]3400.1053410.0703420.1053430.0703440.114***3450.015346Proximity to working zone [level 1]347 − 0.092***3480.033349 − 0.092***3500.033351 − 0.091***3520.005353Proximity to shopping center [level 1]3540.061***3550.0093560.061***3570.0093580.061***3590.003360Proximity to commercial business zone [level 1]361 − 0.0123620.009363 − 0.0123640.009365 − 0.016***3660.001367Proximity to park and green space [level 1]3680.0133690.0113700.0133710.0113720.016***3730.001374Proximity to 3A hospital [level 1]375 − 0.044***3760.003377 − 0.044***3780.003379 − 0.042***3800.001381Proximity to primary and middle school [level 1]382 − 0.084**3830.032384 − 0.084***3850.032386 − 0.088***3870.005388Compactness [level 2]389 − 10.996***3900.327391 − 10.822***3920.054393Centrality [level 2]3941.287***3950.2253961.436***3970.037398Polycentricity [level 2]3990.180***4000.0544010.189***4020.007403Building density [level 2]404 − 0.232***4050.065406 − 0.216***4070.009408Dissimilarity [level 2]4091.458***4100.4254111.682***4120.057413Gini index of population [level 2]414 − 1.375***4150.227416 − 1.457***4170.037418Moran index of education [level 2]419 − 0.718*4200.393421 − 0.835***4220.043423GDP per capita [level 2]42410.905***4251.04242610.894***4270.143428Level 1 Constant42924.5854305.99843111.8494325.63343310.9014345.49443510.1934360.420437Level 2 Constant4380.4324390.1024400.4724410.1124420.0024430.000444N.A445N.A446Level 2 Residual4470.1994480.0244490.1314500.0124510.1314520.012453N.A454N.A455Residual intra − class correlation4560.6834570.7814580.016459N.A460Obs46147,31646247,31646347,31646447,316465Groups466104671046810469N.A470AIC47158,176.6347238,539.5747338,502.1347438,939.39475BIC47658,211.6947738,644.7547838,577.4247939,097.15480481482===== PAGE 17 (score 19, numeric-paren 9, table-words 1, var-words 8) =====4834841003485Multilevel effects of urban form and urban functional zones…4861 3487In the comparison of Models 2 and 3 in Table 5, the residential intra-class correlation reveals 488that the inter-city differences explain 76.5% of the variance of HEPrice. This answers exactly 489how much of the variance of HEPrice is contributed by the city’s uniqueness. Model 3 also 490validates our third hypothesis that housing price variation is partially produced by differentiated 491urban forms and functions. Specifically, the HEPrice would decrease significantly (− 10.996***) 492with the increase of compactness, while centrality is positively associated with the housing 493price (1.287***). Polycentricity stimulates the city-level housing price (0.180***) because 494more working subcenters could attract migrants, thereby increasing the demands of residential 495settlements (Wen & Tao, 2015). The dissimilarity index is positively associated with the city’s 496HEPrice (1.458***), indicating that the uneven land development of the city is associated with 497high housing prices on the overall level. This result supplements the evidence of the impact of 498polycentricity on housing prices from another dimension. Balanced development was achieved 499through the allocation of state investment, which often drives the construction of new satellite 500cities/towns (Liu et al., 2018). The density index negatively affects the HEPrice (− 0.232***), 501suggesting that a high dwelling density is related to a low city-level housing price. This result is 502inconsistent with the previous intra-city evidence of housing premiums induced by local density 503(Kulish et al., 2012). High-density areas at the local level are often urban CBDs where the afflu-504ent are attracted to live nearby, thereby increasing the housing prices. Conversely, for city-level 505density comparison, high urban density typically means a high level of house supply at the over-506all level. When controlling other variables, such as the GDP and the population structure, a large 507amount of supply would depreciate the values of individual houses if the demand is fixed.508In terms of urban functional zones, Model 3 shows that only urban functions of working 509zones (− 0.092***), 3A hospitals (− 0.044***), and primary and middle schools (− 0.084***) 510have premiums on housing prices after controlling the city’s differences. The findings are con-511sistent with most of the previous empirical evidence that working opportunities, convenient com-512muting, and valuable amenities, like hospitals and schools, have capitalization effects on housing 513prices (Wang et al., 2017; Wen et al., 2017). Conversely, proximity to shopping centers likely 514inhibits the surrounding housing values (0.061***). This result is consistent with that of a previ-515ous study where large-scale and high-end shopping complexes have a significant negative impact 516on the housing prices in core urban areas (Zhang et al., 2020).5175 Conclusions518Although previous studies have affirmed the contribution of a city’s uniqueness to the housing 519market in that city, this study supplements the empirical evidence of the influence of urban forms 520and urban functional zones on the differentiated performance of the housing market on multiple 521scales (within and between cities). Our findings indicate that the difference in housing prices 522across cities is not only attributable to the city’s uniqueness in the social and economic environ-523ment (Mirkatouli et al., 2018; Waltert & Schläpfer, 2010; Wang et al., 2017; Yi & Huang, 2014) 524but also to the urban form, which is led by urban growth strategies.525For the different urban forms in the five dimensions, our results show that land use 526development policies for low compactness, high centrality, high polycentricity, low density, 527and high dissimilarity in housing development may significantly increase the average hous-528ing price of the city’s housing market. This phenomenon indicates that paying attention to 529the housing unaffordability within the cities due to the influences of house location should 530also regulate the housing market on the city level through appropriate urban planning and 531design. For example, when considering developing a new urban sub-center, constructing 532533534===== PAGE 18 (score 26, numeric-paren 0, table-words 1, var-words 24) =====5355361004537 538G. Huang et al.5391 3540high-density communities could provide more houses for migrants and reduce the housing 541prices (Fesselmeyer & Seah, 2018). Our research proposes to examine the impacts of the 542compactness, centrality, polycentricity, and dissimilarity in housing development and the 543density on housing prices in the evaluation of urban development projects.544For the distribution of urban functional zones within the cities, housing prices are affected by 545the spatial distribution of urban functions is well studied in previous studies (Qiao et al., 2021; 546Tam et al., 2019; Wang et al., 2017; Wen et al., 2017). But our finding provides a unified per-547spective for such single-city housing price studies (Shen & Karimi, 2017). After controlling the 548city as an influencing factor of housing price differences, working centers, high-quality hospitals, 549and schools can increase the housing prices of neighboring areas.550This research has its limitations. The spatial form of a city is always more or less related 551to its level of economic development. The world’s most economically developed global cit-552ies (e.g., Chicago, Tokyo, Hong Kong, New York, Shanghai) are high-density cities with 553high housing prices. We cannot rashly infer that if a city is actively designed and built into 554a certain type of form, it will automatically push up or decrease local housing prices. The 555focus of this research is to answer whether the design and planning of the city’s physical 556facilities and urban functional zones’ layouts can certainly affect the housing prices if the 557social and economic conditions are equivalent to other cities. The findings emphasize that 558the moderating role of urban design and planning in social inequality caused by the hous-559ing unaffordability is feasible and our conclusions highlight a series of urban design indica-560tors as a toolbox for planners and policymakers.561Appendix562See Tables 6, 7, 8, 9. 563Table 6 POI categorizations of Gaode Map used in this study564Top type (13 categories, all)565Middle type (119 catego-566ries, examples)567Bottom type (2168 categories, examples)568Accommodation service569Hotel, inn570Economical hotel, motel, hostel571Commercial housing572Residential building, 573industrial park574Mansion, apartment, creative industry 575center576Daily life service577Ticket office, express578Train ticket office, barber shop, express579Educational and cultural service School, museum580University, high school, science museum581Financial and insurance service582Bank, insurance583Bank of China, China Pacific Insurance 584Company, ATM585Firms and enterprises586Company, factory587Telecom company, medical company, con-588struction company589Food and beverage590Chinese food, fast food591Sichuan food, fast food, sea food592Government agencies and 593social organizations594Social organization, 595government organization596Charity, court, city hall597Health care service598Hospital, clinic599General hospital, clinic, drug store600Shopping service601Supermarket, mall602Wal-Mart, book store, clothing store603Sports and recreation604Sports, recreation605Fitness center, cinema, playground606Tourist attraction607Scenic spot, park and 608square609Park, temple, botanical garden610Transportation facility611Airport, train station612Airport site, train station, subway station613614615===== PAGE 21 (score 56, numeric-paren 44, table-words 1, var-words 1) =====6166171007618Multilevel effects of urban form and urban functional zones…6191 3620Table 9 Fisher-Hayter pairwise comparisons of HEPrice for ten study cities621City vs. City622City means#623Mean diff624FH-test625Beijing vs. Guangzhou62665.562735.962829.5762996.3627*630Beijing vs. Shanghai63165.563265.16330.426341.6286635Beijing vs. Nanjing63665.563734.063831.4763997.7773*640Beijing vs. Wuhan64165.564216.464349.08644156.6610*645Beijing vs. Chengdu64665.564712.064853.44649202.4649*650Beijing vs. Xi’an65165.565214.165351.39654147.9686*655Beijing vs. Hangzhou65665.565739.265826.2865978.5239*660Beijing vs. Changsha66165.566210.066355.47664139.5118*665Beijing vs. Shenzhen66665.566761.66683.8466911.5009*670Guangzhou vs. Shanghai67135.967265.167329.16674103.4482*675Guangzhou vs. Nanjing67635.967734.06781.896795.5054*680Guangzhou vs. Wuhan68135.968216.468319.5068458.1683*685Guangzhou vs. Chengdu68635.968712.068823.8768982.3705*690Guangzhou vs. Xi’an69135.969214.169321.8269459.3953*695Guangzhou vs. Hangzhou69635.969739.26983.306999.2788*700Guangzhou vs. Changsha70135.970210.070325.8970462.3638*705Guangzhou vs. Shenzhen70635.970761.670825.7470972.6323*710Shanghai vs. Nanjing71165.171234.071331.05714104.1867*715Shanghai vs. Wuhan71665.171716.471848.66719168.5061*720Shanghai vs. Chengdu72165.172212.072353.03724226.2522*725Shanghai vs. Xi’an72665.172714.172850.98729156.6581*730Shanghai vs. Hangzhou73165.173239.273325.8673482.9332*735Shanghai vs. Changsha73665.173710.073855.05739145.4147*740Shanghai vs. Shenzhen74165.174261.67433.4274411.0098*745Nanjing vs. Wuhan74634.074716.474817.6174950.4725*750Nanjing vs. Chengdu75134.075212.075321.9875471.9427*755Nanjing vs. Xi’an75634.075714.175819.9375952.4602*760Nanjing vs. Hangzhou76134.076239.27635.1976414.0845*765Nanjing vs. Changsha76634.076710.076824.0076956.3024*770Nanjing vs. Shenzhen77134.077261.677327.6377475.2126*775Wuhan vs. Chengdu77616.477712.07784.3677914.7141*780Wuhan vs. Xi’an78116.478214.17832.317846.2039*785Wuhan vs. Hangzhou78616.478739.278822.8078963.1849*790Wuhan vs. Changsha79116.479210.07936.3979415.2093*795Wuhan vs. Shenzhen79616.479761.679845.24799125.7092*800Chengdu vs. Xi’an80112.080214.18032.058046.1722*805Chengdu vs. Hangzhou80612.080739.280827.1680985.1580*810Chengdu vs. Changsha81112.081210.08132.028145.2634*815Chengdu vs. Shenzhen81612.081761.681849.60819156.0455*820Xi’an vs. Hangzhou82114.182239.282325.1182464.2650*825Xi’an vs. Changsha82614.182710.08284.078299.1383*830Xi’an vs. Shenzhen83114.183261.683347.55834121.9721*835Hangzhou vs. Changsha83639.283710.083829.1983966.9353*840Hangzhou vs. Shenzhen84139.284261.684322.4484459.2678*845Changsha vs. Shenzhen84610.084761.684851.63849118.6187*850851852===== PAGE 22 (score 8, numeric-paren 0, table-words 1, var-words 6) =====8538541008855 856G. Huang et al.8571 3858Acknowledgements This research project is funded by the National Key R&D Program of China 859(2019YFB1600703).860Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, 861which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long 862as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Com-863mons licence, and indicate if changes were made. The images or other third party material in this article 864are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the 865material. If material is not included in the article’s Creative Commons licence and your intended use is not 866permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly 867from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.868References869Adnan, M., Longley, P. A., Singleton, A. D., & Brunsdon, C. (2010). Towards real-time geodemographics: 870Clustering algorithm performance for large multidimensional spatial databases. Transactions in GIS, 87114(3), 283–297.872Alonso, W. (1964). Location and Land Use. Harvard University Press.873Ashby, D. I., & Longley, P. A. (2005). Geocomputation, geodemographics and resource allocation for local 874policing. Transactions in GIS, 9(1), 53–72.875Batty, M. (2008). The size, scale, and shape of cities. Science, 319(5864), 769–771.876Batty, M., Axhausen, K. W., Giannotti, F., Pozdnoukhov, A., Bazzani, A., Wachowicz, M., Ouzounis, G., & 877Portugali, Y. (2012). Smart cities of the future. The European Physical Journal Special Topics, 214(1), 878481–518.879Batty, M., & Longley, P. A. (1994). Fractal Cities: A Geometry of Form and Function. Academic press.880Cao, R., Tu, W., Yang, C., Li, Q., Liu, J., Zhu, J., Zhang, Q., Li, Q., & Qiu, G. (2020). Deep learning-based 881remote and social sensing data fusion for urban region function recognition. ISPRS Journal of Photo-882grammetry and Remote Sensing, 163, 82–97.883China, N. B. o. S. o. (2020). China Statistical Yearbook 2020. China Statistics Press884Clifton, K., Ewing, R., Knaap, G. J., & Song, Y. (2008). Quantitative analysis of urban form: A multidisci-885plinary review. Journal of Urbanism: International Research on Placemaking and Urban Sustainabil-886ity, 1(1), 17–45. https://doi.org/10.1080/17549170801903496887Crooks, A., Pfoser, D., Jenkins, A., Croitoru, A., Stefanidis, A., Smith, D., Karagiorgou, S., Efentakis, A., 888& Lamprianidis, G. (2015). Crowdsourcing urban form and function. International Journal of Geo-889graphical Information Science, 29(5), 720–741.890Dempsey, N., Brown, C., Raman, S., Porta, S., Jenks, M., Jones, C., & Bramley, G. (2008). Elements of 891Urban Form. (pp. 21–51). Springer Netherlands. https://doi.org/10.1007/978-1-4020-8647-2_2892Dempsey, N., Brown, C., Raman, S., Porta, S., Jenks, M., Jones, C., & Bramley, G. (2010). Elements of 893urban form. In Dimensions of the Sustainable City (pp. 21–51). Springer.894Dong, T., Jiao, L., Xu, G., Yang, L., & Liu, J. (2019). Towards sustainability? Analyzing changing urban 895form patterns in the United States, Europe, and China. Science of the Total Environment, 671, 632–643.896Du, S., Du, S., Liu, B., Zhang, X., & Zheng, Z. (2020). Large-scale urban functional zone mapping by 897integrating remote sensing images and open social data. Giscience & Remote Sensing, 57(3), 411–430.898Duncan, M. (2011). The impact of transit-oriented development on housing prices in San Diego. CA. Urban 899Studies, 48(1), 101–127.900Echenique, M. H., Hargreaves, A. J., Mitchell, G., & Namdeo, A. (2012). Growing cities sustainably: Does 901urban form really matter? Journal of the American Planning Association, 78(2), 121–137.902Feng, X., & Humphreys, B. R. (2012). The impact of professional sports facilities on housing values: Evi-903dence from census block group data. City, Culture and Society, 3(3), 189–200.904Table 9 (continued)905Studentized range critical value (0.05, 9, 47,306) equals to 4.3865448906# the unit of price is 1,000 CNY907* Means the difference is significant between two cities908