1# RECORD 23a821c97160e852a6c2dd5c6097aa79bfac7ed32# TITLE: Prediction and analysis of residential house price using a flexible spatiotemporal model3# YEAR: 2022.0 VENUE: Journal of Applied Economics DOI: 10.1080/15140326.2022.20454664# AUTHORS: Lu Wang; Guang-Xing Wang; Huan Yu; Fei Wang5# PAGES: 22 CANDIDATE TABLE PAGES (1-based): [15]6# ABSTRACT: ABSTRACT House price prediction has traditionally been approached using linear or spatial linear hedonic models and focused on big cities. In this study, we developed a flexible spatiotemporal model (FSTM) to explore the spatiotemporal characteristics of the residential house price and the impact factors in middle-small cities. The FSTM integrated both spatial and temporal components of the residential house price, accounted for its spatiotemporal characteristics, and reproduced its spatial variability and temporal trends. The results showed that the governmental policy had a significant influence on the house price and led to the characteristics being different from those in big cities. The significant factors also included the density of roads, the density of banks, density of supermarkets, the area used by public and user shared area within a building. This study implied that FSTM provided the potential for spatiotemporal prediction of the residential house price in the middle-small cities.789===== PAGE 1 (score 3, numeric-paren 0, table-words 0, var-words 3) =====1011Wang, Lu; Wang, Guangxing; Yu, Huan; Wang, Fei12Article13Prediction and analysis of residential house price using a14flexible spatiotemporal model15Journal of Applied Economics16Provided in Cooperation with:17University of CEMA, Buenos Aires18Suggested Citation: Wang, Lu; Wang, Guangxing; Yu, Huan; Wang, Fei (2022) : Prediction and analysis19of residential house price using a flexible spatiotemporal model, Journal of Applied Economics, ISSN201667-6726, Taylor & Francis, Abingdon, Vol. 25, Iss. 1, pp. 503-522,21https://doi.org/10.1080/15140326.2022.204546622This Version is available at:23https://hdl.handle.net/10419/31417224Standard-Nutzungsbedingungen:25Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen26Zwecken und zum Privatgebrauch gespeichert und kopiert werden.27Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle28Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich29machen, vertreiben oder anderweitig nutzen.30Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen31(insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten,32gelten abweichend von diesen Nutzungsbedingungen die in der dort33genannten Lizenz gewährten Nutzungsrechte.34Terms of use:35Documents in EconStor may be saved and copied for your personal36and scholarly purposes.37You are not to copy documents for public or commercial purposes, to38exhibit the documents publicly, to make them publicly available on the39internet, or to distribute or otherwise use the documents in public.40If the documents have been made available under an Open Content41Licence (especially Creative Commons Licences), you may exercise42further usage rights as specified in the indicated licence.43 https://creativecommons.org/licenses/by/4.0/444546===== PAGE 15 (score 29, numeric-paren 0, table-words 7, var-words 9) =====4748to the year 2004, although their p-values were larger than the significant level of 0.05, 49indicating a statistical insignificance. This implied that the decrease of the residential 50house price was slight and not significant. Starting in 2005, the coefficients became 51positive, increased over years, and showed statistically significant effect on the residential 52house price at the significant level of 0.05 except the year 2006 in which the p-value was 530.054, slightly smaller than 0.05. This indicated that overall, the increase of the residential 54house price over the years was statistically significant.55For the factors that drove the spatial variability of the residential house price, there 56were only five factors that had the p-values smaller than 0.05, including DensityRoad, 57DensitySupmarket, DensityBank, AreaUsedP and AreaShareAreaB. DensityRoad with 58a negative coefficient significantly affected the residential house price, meaning that the 59higher the density of roads, the lower the residential house price. Within a city, the higher 60road density often implies an old downtown where there is much noise and the houses 61were old and had lower quality, which would reduce one’s desirability of buying houses. 62DensitySupmarket also had a significant negative effect on the residential house price. In 63addition to noise, super markets often cause safe problems. Moreover, the residential 64houses are usually far away from super markets. Banks (DensityBank) provided conve65nience for residents and thus had significantly positive an effect on the residential house 66price. The areas used by public (AreaUsedP) provided facilities such as recreation and 67sport areas for the residents and thus enhanced the extraction for the residents to buy the 68houses. The shared areas within buildings had a positive coefficient, seemed not reason69able because the residents paid more for the residents shared areas. In fact, the residents 70liked to live in the buildings that had good facilities such as elevators and sport rooms.71Table 1. The coefficients of flexible spatiotemporal model based on the 72sample data of residential house price.73Coefficient74Standard error75p-value76(Intercept)772974.6678222.83790.00080as.factor(Time)200281−120.3082277.37830.66584as.factor(Time)200385−164.3786260.14870.52888as.factor(Time)200489−456.4890290.14910.11692as.factor(Time)200593417.3894296.57950.16096as.factor(Time)200697575.5398298.60990.054100as.factor(Time)2007101792.12102316.111030.012104as.factor(Time)20081051723.92106336.211070.000108as.factor(Time)20091092172.30110321.501110.000112as.factor(Time)20101133341.63114406.181150.000116DisCenter117−130.66118135.561190.335120DensityRoad121−411.78122135.611230.002124DensityMarket12588.65126140.721270.529128DensityBusS129−108.46130111.141310.329132DensityTeleCom133−107.58134153.381350.483136DensitySupmarket137−219.79138111.021390.048140DensityPost141−57.8214299.851430.563144DensityBank145423.41146197.041470.032148AreaUsedP149234.2015078.761510.003152UserShareAreaB153267.5115479.711550.001156Density_2003Pop15779.39158112.921590.482160JOURNAL OF APPLIED ECONOMICS161515162163164===== PAGE 16 (score 25, numeric-paren 0, table-words 9, var-words 7) =====165166Although other spatially driving factors did not significantly affect the residential 167house price, their coefficients were meaningful. DisCenter had a negative coefficient, 168suggesting that the larger the distances to the city center, the smaller the residential house 169price values. DensityBusS, DensityTeleCom and DensityPost all had negative coefficients 170because of noise and safety reasons. Moreover, the population densities of all the years 171were involved in the FSTM but removed except the population density of the year 2003 172with a positive coefficient.1733.4. The spatiotemporal predictions and accuracy assessment using FSTM174The obtained FSTM was used to make the spatiotemporal predictions of the residential 175house price that consisted of the spatial distribution maps of the residential house price 176for the years from 2001 to 2010. The 10-fold cross validation led to a RMSE value of 2087 177Chinese Yuan/m2, a relative RMSE of 58.8% (Table 2), a RMSPE value of 0.568 and 178a MAPE value of 0.41 (Table 3). The RMSPE and MAPE were statistically slightly 179significantly larger than zero at the significant level of 0.05 (Table 3).180In Table 2, the large RMSE and relative RMSE values were caused mainly by the large 181overall coefficient of variation, 86.4%. The annual relative RMSE ranged from 31.9% to 182114.2%. The extremely large value of the annual relative RMSE, 114.2%, occurred in 183the year of 2005 mainly because in 2005 there was one extremely large house price, 56,616 184Chinese Yuan/m2, collected from a business center, while the annual sample mean of 185the year 2005 was 3377 Chinese Yuan/m2. When developing the FSTM, we did not 186Table 2. The accuracy assessment of predicted residential house price values from 187FSTM based on root mean square error (RMSE) and relative RMSE (=RMSE×100/ 188sample mean).189Year190Sample mean191Average estimate192RMSE193RRMSE194200119528251962825197172619861.1199200220027632012763202178420364.6204200320526952062695207147220854.6209200421024682112468212128821352.22142005215337721633772173857218114.2219200622037102213710222154622341.7224200722541702264170227148822835.7229200823050552315055232295023358.3234200923554422365442237201023836.9239201024067272416598242214724331.9244Overall24535512463551247208724858.8249Table 3. Comparison of performance results for models as a spatial predictor (RMSPE and MAPE).250RMSPE MAPE R Square251Cross 252Validation 253Method254Independent Variable Set255FSM2560.5682570.4102580.27525910 fold cross 260validation261PricePer ~ List((Time)+ (DisCenter+DensityRoad+DensityMarket 262+DensityCar+DensityCom+DensitySupmarket+DensityPost 263+DensityBank+LandComArea+LandShareArea 264+Density_2003Pop)265516266L. WANG ET AL.267