# Spatial Determinants of Real Estate Appraisals in The Netherlands: A Machine Learning Approach

^{1}

^{2}

^{3}

^{4}

^{*}

## Abstract

**:**

## 1. Introduction

- Which ML approaches are currently used for hedonic pricing, and how do they perform?
- Which factors are significant for price differences between houses across cities?
- Which data are available about these factors?
- How can we construct a method for hedonic pricing across different cities using the obtained insights?
- What are the results of applying this method with a realistic dataset?

## 2. Background

#### 2.1. Dutch House Price Indices and the Repeat-Sales Model

#### 2.2. Hedonic Price Models

#### 2.3. Linear Regression (LR)

#### 2.4. Geographically Weighted Regression (GWR)

#### 2.5. Multi-Scale Geographically Weighted Regression (MGWR)

#### 2.6. Regression Trees and Extreme Gradient Boost (XGBoost)

#### 2.7. Features for House Price Estimations

## 3. Data and Methods

#### 3.1. Model Metrics

#### 3.1.1. Quantitative Metrics

#### 3.1.2. Qualitative Metrics

#### 3.2. Exploration of the Response Variable

#### 3.3. Exploration of the Explanatory Variables

#### 3.4. Hyper-Parameter Optimisation Using CV

## 4. Results

## 5. Discussion

## 6. Conclusions

Which ML approaches are currently used for hedonic pricing, and how do they perform?

Which factors are significant for price differences between houses across cities? Which data are available about these factors?

How can we construct a method for hedonic pricing across different cities using the obtained insights? What are the results of applying this method with a realistic dataset?

- -
- The lack of a feature to model house quality. The remaining unexplained variance of 17% is likely due to a missing variable that explains the quality of the house itself or other location characteristics. An official appraisal report contains more detailed information about the state of a house. This can help paint a better picture of the house itself.
- -
- For example, the ground sinkage map from TU Delft provides an interesting use case for looking at real estate portfolio risk factors. Ground sinkage is a real problem in the Netherlands, especially in Groningen. As a result of the gas exploitation, the property values are reduced drastically in the region. This poses a clear risk to the mortgage owner and the money lender. Another problem for many houses is foundation rot; perhaps risk areas can be identified by combining sinkage data with ground compositions.

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## Abbreviations

LR | Linear regression |

(M)GWR | (Multi-scale) Geographically weighted regression |

XGBoost | Extreme gradient boosting |

CBS | ‘Centraal Bureau voor de Statistiek’ (ENG: Central Agency for Statistics) |

BAG | ‘Basisregistratie adressen & gebouwen’ (ENG: Base registry addresses & buildings) |

DKK | ‘Digitale kadastrale kaart’ (ENG: Digital cadastral map) |

## Appendix A. Figures Related to Dutch Housing Market

**Table A1.**% change in house prices (January 2000–January 2020), per provinces of the Netherlands [42].

Province | % Increase over 2000–2020 | Province | % Increase over 2000–2020 |
---|---|---|---|

Drenthe | 56.35% | Noord-Brabant | 42.90% |

Flevoland | 44.02% | Noord-Holland | 76.70% |

Friesland | 55.34% | Overijssel | 49.73% |

Gelderland | 45.05% | Utrecht | 70.87% |

Groningen | 67.48% | Zeeland | 74.83% |

Limburg | 38.16% | Zuid-Holland | 52.36% |

**Table A2.**% change in house prices (January 2000–January 2020), per housing type. Source: Kadaster [42].

Housing Type | % Increase over 2000–2020 |
---|---|

Detached | 54.4% |

Semi-detached | 51.2% |

Terraced House | 64.0% |

Corner House | 61.5% |

Apartment | 75.3% |

## Appendix B. Figures Related to Models

**Figure A1.**Number of real estate appraisals of Stater, (

**left**) 2008, (

**middle**) 2020, (

**right**) January 2000–January 2021.

**Figure A2.**Exploration of external variables from Kadaster & CBS (Amersfoort, 2018). (

**a**) Kadaster—Land lot size (${\mathrm{m}}^{2}$) & total floor area (${\mathrm{m}}^{2}$). (

**b**) RVO—Energy Labels.

Municipality | Kernel (Bandwidth) | Adaptive/Fixed |
---|---|---|

Amersfoort | Gaussian (0.28) | Adaptive |

Amsterdam | Gaussian (0.19) | Adaptive |

Eindhoven | Gaussian (0.27) | Adaptive |

Groningen | Gaussian (0.43) | Adaptive |

Rotterdam | Gaussian (0.25) | Adaptive |

Abbreviation | Description | VIF | Source |
---|---|---|---|

is_gezinwng | Apartment (0) or Family home (1) | 1.91 | Stater |

garage | Presence of garage (yes/no = 1/0) | 1.82 | - |

parkeerplaats | % of people aged 0–14 (500 m tiles) | 1.43 | - |

vbo_oppervlakte | The total floor area of a house (${\mathrm{m}}^{2}$) | 1.82 | Kadaster |

pnd_bouwjaar | Build year | 2.31 | - |

perceel_oppr | The total floor area of a house (${\mathrm{m}}^{2}$) | 1.80 | - |

Pand_energieklasse | Energy label / class (factor) | 2.69 | RVO |

INW_014 | % of people aged 0–14 (500 m tiles) | 15.05 | CBS |

INW_1524 | % of people aged 15–24 (500 m tiles) | 1.53 | - |

INW_2544 | % of people aged 25–44 (500 m tiles) | 11.47 | - |

INW_4464 | % of people aged 45–64 (500 m tiles) | 7.67 | - |

INW_65PL | % of people aged 65+ (500 m tiles) | 10.91 | - |

TOTHH_EENP | % of single person house holds | 4.12 | - |

TOTHH_MPZK | % of households > 1 and no children | 4.78 | - |

HH_EENOUD | % of one parent households with children | 4.68 | - |

WON_MRGEZ | % of family homes | 4.4 | - |

WON_NBEW | % non-inhabited homes | 1.90 | - |

OAD | Address density (address/${\mathrm{km}}^{2}$) | 3.87 | - |

STED_500 | Urbanisation (factor) | 6.12 | - |

P_KOOPWON | % owner-occupied home | 3.43 | - |

WOZWONING | Average WOZ-Waarde (×1000€) | 3.15 | - |

M_INKHH | Median income group (factor) | 4.12 | - |

G_ELEK_WON | Average Electricity Usage (kwH) | 2.10 | - |

P_LINK_HH | % of households belonging to bottom 40% of national income | 13.12 | - |

P_HINK_HH | % of households belonging to top 20% of national income | 14.45 | - |

AFS_SUPERM | Distance to nearest supermarket (km) | 3.22 | - |

AFS_OPRIT | Distance to nearest provincial road or highway (km) | 2.48 | - |

AFS_CAFE | Distance to nearest cafe (km) | 2.21 | - |

AFS_BIBLIO | Distance to nearest library (km) | 2.27 | - |

AFS_ONDVRT | Distance to nearest secondary education (km) | 1.77 | - |

AFS_APOTH | Distance to nearest pharmacy (km) | 2.08 | - |

Abbreviation | Description | Source |
---|---|---|

dist_centre | Distance to city center (km) | Self-computed |

UITKMINAOW | Income from state pension (AOW) | CBS |

INWONER | Inhabitants at start of year | - |

AANTAL_HH | Number of households. | - |

HH_TWEEOUD | % of two parent households with children | - |

P_NW_MIG_A | Percentage of inhabitants (non-western) | - |

P_HUURWON | Percentage of rented homes | - |

G_GA_WON | Average Gas Usage (${\mathrm{m}}^{3}$) | - |

AV1/5/10/20 vars. | Variables describing ’Amount of X within radius 1/5/10/20 km’ (hospitals, stores, schools etc.) | - |

Other AFS vars. | Distance variables to other amenities (Swimming pool, attraction parks, restaurants, hotels, hospital and others.) | - |

Pand_gebouwtype | Home type | RVO |

Pand_subtype | Home subtype | - |

**Figure A3.**Variable importance for the LR model of Amersfoort (2018). All 5 municipalities have similar results.

Municipality | ${\mathit{R}}^{2}$ | RMSE | MAE | MAPE |
---|---|---|---|---|

Amersfoort | 0.810 | €61,928 | €50,177 | 7.51% |

Amsterdam | 0.822 | €62,596 | €52,183 | 7.40% |

Eindhoven | 0.815 | €62,942 | €54,631 | 7.98% |

Groningen | 0.821 | €79,192 | €54,131 | 8.29% |

Rotterdam | 0.837 | €58,561 | €49,287 | 7.25% |

**Figure A4.**Model fit of XGBoost models for Amsterdam, Eindhoven, Rotterdam, Groniningen (2018), (orange line is y = x).

**Figure A5.**XGBoost Variable Importance of Amersfoort & Amsterdam (2018). (

**a**) Amersfoort. (

**b**) Amsterdam.

## References

- AFM. Hypotheek in Relatie tot Waarde Huis (LTV). Available online: https://afm.nl/nl-nl/consumenten/themas/producten/hypotheek/hoeveel-lenen/maximale-hypotheek (accessed on 2 February 2021).
- Lozej, M.; Rannenberg, A. The macroeconomic effects of the LTV and LTI ratios in Ireland. Appl. Econ. Lett.
**2018**, 25, 1507–1511. [Google Scholar] [CrossRef] - De Nederlandsche Bank. The Quality and Integrity of Residential Real Estate Appraisals; Occasional Studies; De Nederlandsche Bank: Amsterdam, The Netherlands, 2019. [Google Scholar]
- Brander, L.M.; Koetse, M.J. The value of urban open space: Meta-analyses of contingent valuation and hedonic pricing results. J. Environ. Manag.
**2011**, 92, 2763–2773. [Google Scholar] [CrossRef] [PubMed] - Potrawa, T. Hedonic Pricing Model for Rotterdam Housing Market. Master’s Thesis, TU Delft, Delft, The Netherlands, 2020. [Google Scholar]
- Liebelt, V.; Bartke, S.; Schwarz, N. Hedonic pricing analysis of the influence of urban green spaces onto residential prices: The case of Leipzig, Germany. Eur. Plan. Stud.
**2018**, 26, 133–157. [Google Scholar] [CrossRef] - Cao, K.; Diao, M.; Wu, B. A big data–based geographically weighted regression model for public housing prices: A case study in Singapore. Ann. Am. Assoc. Geogr.
**2019**, 109, 173–186. [Google Scholar] [CrossRef] - Waarderingskamer. Hoe de WOZ-Waarde tot Stand Komt. Available online: https://waarderingskamer.nl/klopt-mijn-woz-waarde/totstandkoming-woz-waarde/ (accessed on 1 March 2021).
- Kadaster. BAG, Addresses and Buildings Key Register. Available online: https://kadaster.nl/zakelijk/registraties/basisregistraties/bag (accessed on 15 January 2021).
- Calcasa. WOX-Waarde. Available online: https://calcasa.nl/wox-online (accessed on 2 March 2021).
- Jansen, S.; de Vries, P.; Coolen, H.; Lamain, C.; Boelhouwer, P. Developing a house price index for the Netherlands: A practical application of weighted repeat sales. J. Real Estate Financ. Econ.
**2008**, 37, 163–186. [Google Scholar] [CrossRef] [Green Version] - Harding, J.P.; Rosenthal, S.S.; Sirmans, C. Depreciation of housing capital, maintenance, and house price inflation: Estimates from a repeat sales model. J. Urban Econ.
**2007**, 61, 193–217. [Google Scholar] [CrossRef] - Goodman, A.C. Hedonic prices, price indices and housing markets. J. Urban Econ.
**1978**, 5, 471–484. [Google Scholar] [CrossRef] - Luttik, J. The value of trees, water and open space as reflected by house prices in the Netherlands. Landsc. Urban Plan.
**2000**, 48, 161–167. [Google Scholar] [CrossRef] - Farber, S.; Yeates, M. A comparison of localized regression models in a hedonic house price context. Can. J. Reg. Sci.
**2006**, 29, 405–420. [Google Scholar] - Zhou, J.; Zhang, H.; Gu, Y.; Pantelous, A.A. Affordable levels of house prices using fuzzy linear regression analysis: The case of Shanghai. Soft Comput.
**2018**, 22, 5407–5418. [Google Scholar] [CrossRef] - CBS. Prijzen Koopwoningen. Available online: https://cbs.nl/nl-nl/reeksen/prijzen-koopwoningen (accessed on 16 January 2021).
- Fotheringham, A.; Brunsdon, C.; Charlton, M. Geographically Weighted Regression: The Analysis of Spatially Varying Relationships; John Wiley & Sons: Hoboken, NJ, USA, 2002; Volume 13. [Google Scholar] [CrossRef]
- Gong, Y. The Spatial Dimension of House Prices. A+ BE|Archit. Built Environ.
**2017**, 4, 1–186. [Google Scholar] - McMillen, D.P. The return of centralization to Chicago: Using repeat sales to identify changes in house price distance gradients. Reg. Sci. Urban Econ.
**2003**, 33, 287–304. [Google Scholar] [CrossRef] - Tomal, M. Modelling Housing Rents Using Spatial Autoregressive Geographically Weighted Regression: A Case Study in Cracow, Poland. ISPRS Int. J. Geo-Inf.
**2020**, 9, 346. [Google Scholar] [CrossRef] - Tobler, W.R. A computer movie simulating urban growth in the Detroit region. Econ. Geogr.
**1970**, 46, 234–240. [Google Scholar] [CrossRef] - Sims, S.; Dent, P.; Oskrochi, G.R. Modelling the impact of wind farms on house prices in the UK. Int. J. Strateg. Prop. Manag.
**2008**, 12, 251–269. [Google Scholar] [CrossRef] [Green Version] - Wu, C.; Ren, F.; Hu, W.; Du, Q. Multiscale geographically and temporally weighted regression: Exploring the spatiotemporal determinants of housing prices. Int. J. Geogr. Inf. Sci.
**2018**, 33, 489–511. [Google Scholar] [CrossRef] - Zhang, S.; Wang, L.; Lu, F. Exploring housing rent by mixed geographically weighted regression: A Case study in Nanjing. ISPRS Int. J. Geo-Inf.
**2019**, 8, 431. [Google Scholar] [CrossRef] [Green Version] - Shabrina, Z.; Buyuklieva, B.; Ng, M.K.M. Short-Term Rental Platform in the Urban Tourism Context: A Geographically Weighted Regression (GWR) and a Multiscale GWR (MGWR) Approaches. Geogr. Anal.
**2021**, 53, 686–707. [Google Scholar] [CrossRef] - Shahhosseini, M.; Hu, G.; Pham, H. Optimizing Ensemble Weights for Machine Learning Models: A Case Study for Housing Price Prediction. In Smart Service Systems, Operations Management, and Analytics; Yang, H., Qiu, R., Chen, W., Eds.; Springer International Publishing: Cham, Switzerland, 2020; pp. 87–97. [Google Scholar]
- Avanijaa, J. Prediction of House Price Using XGBoost Regression Algorithm. Turk. J. Comput. Math. Educ. (TURCOMAT)
**2021**, 12, 2151–2155. [Google Scholar] [CrossRef] - So, H.M.; Tse, R.Y.; Ganesan, S. Estimating the influence of transport on house prices: Evidence from Hong Kong. J. Prop. Valuat. Invest.
**1997**, 15, 40–47. [Google Scholar] [CrossRef] - Amri, S.; Tularam, G.A. Performance of multiple linear regression and nonlinear neural networks and fuzzy logic techniques in modelling house prices. J. Math. Stat.
**2012**, 8, 419–434. [Google Scholar] [CrossRef] [Green Version] - Brunsdon, C.; Corcoran, J.; Higgs, G. Visualising space and time in crime patterns: A comparison of methods. Comput. Environ. Urban Syst.
**2007**, 31, 52–75. [Google Scholar] [CrossRef] - de Wit, E.R.; Englund, P.; Francke, M.K. Price and transaction volume in the Dutch housing market. Reg. Sci. Urban Econ.
**2013**, 43, 220–241. [Google Scholar] [CrossRef] [Green Version] - Wheeler, D.C.; Páez, A. Geographically Weighted Regression. In Handbook of Applied Spatial Analysis: Software Tools, Methods and Applications; Springer: Berlin/Heidelberg, Germany, 2010; pp. 461–486. [Google Scholar] [CrossRef]
- McCluskey, J.J.; Rausser, G.C. Hazardous waste sites and housing appreciation rates. J. Environ. Econ. Manag.
**2003**, 45, 166–176. [Google Scholar] [CrossRef] [Green Version] - Ambrose, B.; Eichholtz, P.; Lindenthal, T. House Prices and Fundamentals: 355 Years of Evidence. J. Money Credit. Bank.
**2012**, 45, 477–491. [Google Scholar] [CrossRef] - Fuerst, F.; McAllister, P.; Nanda, A.; Wyatt, P. Does energy efficiency matter to home-buyers? An investigation of EPC ratings and transaction prices in England. Energy Econ.
**2015**, 48, 145–156. [Google Scholar] [CrossRef] [Green Version] - de Groot, H.; de Vor, F. The impact of industrial sites on residential property values: A hedonic pricing analysis from the Netherlands. Reg. Stud.
**2011**, 45, 609–623. [Google Scholar] [CrossRef] [Green Version] - Kadaster/ESRI Nederland. DKK: Land Lot Dataset. Available online: https://arcgis.com/home/group.html?id=eb452ccc59e0431c8b42b06c7e7a6fee#overview (accessed on 5 March 2021).
- CBS. 100 × 100 Squaremeter Statistics. Available online: https://cbs.nl/nl-nl/dossier/nederland-regionaal/geografische-data/kaart-van-100-meter-bij-100-meter-met-statistieken (accessed on 1 March 2021).
- RVO. Energy Labels Dataset. Available online: https://www.ep-online.nl/ (accessed on 1 March 2021).
- Fotheringham, A.; Yang, W.; Kang, W. Multiscale Geographically Weighted Regression (MGWR). Ann. Am. Assoc. Geogr.
**2017**, 107, 1247–1265. [Google Scholar] [CrossRef] - Kadaster. Vastgoed Dashboard, Prijsindex. Available online: https://kadaster.nl/zakelijk/vastgoedinformatie/vastgoedcijfers/vastgoeddashboard/prijsindex (accessed on 15 February 2021).

**Figure 1.**Exploration of the residential real estate appraisal dataset of Stater N.V. (

**a**) Appraisals per year (2000–2020). (

**b**) Records per municipality (2020). (

**c**) Increase in average appraisal value, (Amersfoort, 2000 & 2020).

**Figure 2.**Various CBS 100 × 100 m statistics (Amersfoort, 2018). (

**a**) Taxation value (WOZ-waarde) (€1k). (

**b**) Electricity usage (kWh). (

**c**) Nearest cafe (km).

**Figure 4.**Q—Q plot showing impact on overall fit for including all appraisals (Amersfoort, 2018). (

**a**) All appraisals, poor fit. (

**b**) Appraisals < €750,000, adequate fit.

**Figure 5.**Plots describing the GWR model (Amersfoort, 2018). (

**a**) The influence of living area. (

**b**) Variable importance.

**Figure 7.**Differences between XGBoost prediction and indexation using a regional price index (green = XGBoost predicts higher). (

**a**) For apartments, XGBoost predicts 17.31% higher. (

**b**) For family homes, XGBoost predicts 11.12% higher.

Characteristic | Influence | Sources |
---|---|---|

Year of construction | Positive/Negative | [5,16,34] |

Living area | Strongly positive | [5,13,16] |

Type of housing | Positive | [5,13,16] |

Garden space/presence of garden | Positive | [13,16] |

# of rooms (bedrooms, bathrooms) | Positive | [13,16] |

Presence of facilities (shower, lift, garage, etc.) | Slightly positive | [13,16] |

Furnished | Slightly positive | [13,16] |

Energy Efficiency | Slightly positive | [5] |

Sustainability measures | Slightly positive | [5] |

Characteristic | Influence | Sources |
---|---|---|

Household income | Strongly positive | [7,18] |

House shortage | Strongly positive | [35] |

Notable view (sea, lake, park) | Strongly positive | [33] |

Time to travel or distance to city centre | Strongly positive | [14,19] |

Proximity to place of worship | Positive/Negative | [5,36] |

Distance to highway | Negative | [37] |

Distance to heavy industry | Negative | [37] |

Presence to high rise/view obstruction | Negative | [16] |

Crime rate | Negative | [19] |

Unemployment rate | Slightly negative | [18] |

Population density | Positive | [35] |

Presence of cultural landmarks | Slightly positive | [18] |

Birth surplus | None | [36] |

Municipalities | Samples | Mean | Std Dev. | Min | Max |
---|---|---|---|---|---|

Amersfoort | 1494 | €319,400 | €62,744 | €58,800 | €1,250,000 |

Amsterdam | 5084 | €451,650 | €84,992 | €81,000 | €1,500,000 |

Eindhoven | 1845 | €278,800 | €58,421 | €75,000 | €1,155,000 |

Groningen | 1160 | €222,610 | €49,143 | €45,000 | €955,000 |

Rotterdam | 3011 | €254,930 | €53,329 | €55,000 | €875,000 |

Dataset Name | Contents | Joined Using | Source |
---|---|---|---|

BAG: ‘Addresses and Buildings key register’ | Geo-coordinates, build year, surface area | Address | Kadaster [9] |

DKK: ‘Digital cadastral map’ | Land lot area | BAG-VBO-ID | Kadaster [38] |

CBS Square statistics | Variables for areas of 100 × 100 m and 500 × 500 m | Geo-coordinates | CBS [39] |

EP-Online | Energy labels | BAG-VBO-ID | RVO [40] |

Municipality | Build Year | Land Lot Area | Address Density | Households | Energy Usage | Distance | Energy Label |
---|---|---|---|---|---|---|---|

Amersfoort | 4 (0.27%) | 451 (30.19%) | 15 (1.00%) | 16 (1.07%) | 28 (1.87%) | 15 (1.00%) | 454 (30.39%) |

Amsterdam | 116 (2.28%) | 731 (14.38%) | 0 | 71 (1.40%) | 127 (2.50%) | 0 | 1391 (27.36%) |

Eindhoven | 16 (0.87%) | 659 (35.72%) | 0 | 93 (5.04%) | 2 (0.11%) | 2 (0.11%) | 587 (31.82%) |

Groningen | 25 (2.16%) | 382 (32.93%) | 0 | 97 (8.36%) | 15 (1.29%) | 2 (0.17%) | 312 (26.90%) |

Rotterdam | 1 (0.03%) | 732 (24.31%) | 0 | 39 (1.30%) | 17 (0.56%) | 0 | 948 (31.48%) |

Municipality | WOZ-Waarde | Income |
---|---|---|

Amersfoort | 96 (6.43%) | 74 (4.95%) |

Amsterdam | 398 (7.83%) | 259 (5.09%) |

Eindhoven | 171 (9.27%) | 88 (4.77%) |

Groningen | 138 (11.90%) | 84 (7.24%) |

Rotterdam | 216 (7.17%) | 101 (3.35%) |

Metric | ${\mathit{R}}^{2}$ | RMSE | MAE | MAPE |
---|---|---|---|---|

LR (all appraisals) | 0.709 | €150,211 | €72,391 | 11.81% |

LR * | 0.785 | €85,628 | €56,219 | 9.61% |

LR-LOG * | 0.768 | €89,136 | €63,577 | 10.62% |

Municipality | ${\mathit{R}}^{2}$ | RMSE | MAE | MAPE |
---|---|---|---|---|

Amersfoort | 0.822 | €61,459 | €48,393 | 7.42% |

Amsterdam | 0.831 | €60,213 | €53,671 | 7.31% |

Eindhoven | 0.812 | €62,942 | €54,103 | 8.01% |

Groningen | 0.789 | €83,233 | €55,213 | 8.61% |

Rotterdam | 0.861 | €56,431 | €47,312 | 6.99% |

Municipality | ${\mathit{R}}^{2}$ | RMSE | MAE | MAPE |
---|---|---|---|---|

Amersfoort | 0.851 | €57,391 | €34,283 | 5.38% |

Amsterdam | 0.845 | €57,964 | €35,258 | 5.50% |

Eindhoven | 0.838 | €57,385 | €36,192 | 5.62% |

Groningen | 0.829 | €59,832 | €38,241 | 5.88% |

Rotterdam | 0.871 | €56,144 | €34,831 | 5.45% |

Year | 2018 | 2020 | ||||||
---|---|---|---|---|---|---|---|---|

${\mathit{R}}^{2}$ | RMSE | MAE | MAPE | ${\mathit{R}}^{2}$ | RMSE | MAE | MAPE | |

LR | 0.725 | €97,232 | €67,814 | 10.55% | 0.734 | €94,927 | €62,871 | 10.23% |

GWR | 0.822 | €64,856 | €51,738 | 7.67% | 0.809 | €65,826 | €52,237 | 7.92% |

XGBoost | 0.848 | €58,374 | €35,761 | 5.89% | 0.852 | €61,028 | €35,451 | 5.76% |

${\mathit{R}}^{2}$ | RMSE | MAE | MAPE | |
---|---|---|---|---|

XGBoost | 0.832 | 65,312 | 43,625 | 6.35% |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Guliker, E.; Folmer, E.; van Sinderen, M.
Spatial Determinants of Real Estate Appraisals in The Netherlands: A Machine Learning Approach. *ISPRS Int. J. Geo-Inf.* **2022**, *11*, 125.
https://doi.org/10.3390/ijgi11020125

**AMA Style**

Guliker E, Folmer E, van Sinderen M.
Spatial Determinants of Real Estate Appraisals in The Netherlands: A Machine Learning Approach. *ISPRS International Journal of Geo-Information*. 2022; 11(2):125.
https://doi.org/10.3390/ijgi11020125

**Chicago/Turabian Style**

Guliker, Evert, Erwin Folmer, and Marten van Sinderen.
2022. "Spatial Determinants of Real Estate Appraisals in The Netherlands: A Machine Learning Approach" *ISPRS International Journal of Geo-Information* 11, no. 2: 125.
https://doi.org/10.3390/ijgi11020125