# RECORD 53d1b6d48e64b056e88ebabd287c3f06ec3b559d
# TITLE: Comparing automated valuation models for real estate assessment in the Santiago Metropolitan Region: A study on machine learning algorithms and hedonic pricing with spatial adjustments.
# YEAR: 2025.0  VENUE: PLoS ONE  DOI: 10.1371/journal.pone.0318701
# AUTHORS: Jocelyn Tapia; Nicolas Chavez-Garzon; R. Pezoa; Paulina Suarez-Aldunate; M. Pilleux
# PAGES: 27  CANDIDATE TABLE PAGES (1-based): [11, 20]
# ABSTRACT: This study compares the precision and interpretability of two automated valuation models for evaluating the real estate market in the Santiago Metropolitan Region of Chile: machine learning algorithms, specifically LightGBM, and hedonic prices with spatial adjustments (SAR). Traditional residence attributes, such as housing amenities and proximity to services, were considered alongside visual information extracted from images using Convolutional Neural Networks (CNN). The research evaluates the influence of each model characteristic on performance metrics and identifies the relative importance of attributes using the SHapley Additive exPlanations (SHAP) algorithm. The results demonstrate the positive impact of image-based variables on performance metrics, showing that the introduction of visual information can considerably reduce error margins when estimating housing prices. In addition, the SHAP algorithm reveals complex non-linear interactions between price and crucial variables such as total surface area and neighborhood attributes, highlighting the importance of using methods that can capture these effects. Likewise, both LightGBM and SAR models indicate that variables that have the most significant impact on the value of properties are total surface area, municipality quality index, average academic level of nearby schools, and the number of bathrooms.


===== PAGE 1 (score 11, numeric-paren 0, table-words 0, var-words 11) =====

PLOS ONE
OPEN ACCESS
Citation: Tapia J, Chavez-Garzon N, Pezoa R,
Suarez-Aldunate P, Pilleux M (2025)
Comparing automated valuation models for real
estate assessment in the Santiago Metropolitan
Region: A study on machine learning algorithms
and hedonic pricing with spatial adjustments.
PLoS ONE 20(3): e0318701. https://doi.org/10.
1371/journal.pone.0318701
Editor: Marco Civera, Politecnico di Torino,
ITALY
Received: March 27, 2024
Accepted: January 20, 2025
Published: March 25, 2025
Copyright: © 2025 Tapia 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
and code are included in our submission,
available on Mendeley Data.
Funding: The author(s) received no specific
funding for this work.
Competing interests: The authors have
declared that no competing interests exist.
RESEARCH ARTICLE
Comparing automated valuation models
for real estate assessment in the Santiago
Metropolitan Region: A study on machine
learning algorithms and hedonic pricing
with spatial adjustments
Jocelyn Tapia
 
 
1☯∗, Nicolas Chavez-Garzon1☯, Raúl Pezoa2☯,
Paulina Suarez-Aldunate3¤, Mauricio Pilleux3
1 Department of Business Engineering, Universidad Técnica Federico Santa María, Santiago, Chile,
2 School of Industrial Engineering, Faculty of Engineering and Sciences, Universidad Diego Portales,
Santiago, Chile, 3 Coderhub SpA, Santiago, Chile
¤ PS-A and MP also contributed equally to this work.
☯These authors contributed equally to this work.
∗jocelyn.tapias@usm.cl
Abstract
This study compares the precision and interpretability of two automated valuation mod-
els for evaluating the real estate market in the Santiago Metropolitan Region of Chile:
machine learning algorithms, specifically LightGBM, and hedonic prices with spatial
adjustments (SAR). Traditional residence attributes, such as housing amenities and prox-
imity to services, were considered alongside visual information extracted from images
using Convolutional Neural Networks (CNN). The research evaluates the influence of
each model characteristic on performance metrics and identifies the relative importance
of attributes using the SHapley Additive exPlanations (SHAP) algorithm. The results
demonstrate the positive impact of image-based variables on performance metrics,
showing that the introduction of visual information can considerably reduce error mar-
gins when estimating housing prices. In addition, the SHAP algorithm reveals complex
non-linear interactions between price and crucial variables such as total surface area and
neighborhood attributes, highlighting the importance of using methods that can capture
these effects. Likewise, both LightGBM and SAR models indicate that variables that have
the most significant impact on the value of properties are total surface area, municipality
quality index, average academic level of nearby schools, and the number of bathrooms.
Introduction
The traditional property valuation method is based on the visual and comparative charac-
teristics that an appraiser can gather, where variables such as the year of construction, total
surface area, number of bathrooms and number of bedrooms, among others, are analyzed.
Then, a detailed report is created that indicates the estimated value of the property based on
PLOS ONE https://doi.org/10.1371/journal.pone.0318701
March 25, 2025
1/ 27


===== PAGE 11 (score 39, numeric-paren 0, table-words 11, var-words 13) =====

PLOS ONE
A comparative analysis of automated valuation models: Machine learning and spatially-adjusted hedonic pricing
Table 2. Metrics to measure the performance of the model.
Metric
Abbreviation
Formula
Mean absolute percentage error
MAPE
100
N ∑N
i=1 | yi– ̂yi
yi |
Median absolute percentage error
MdAPE
med| yi– ̂yi
yi | ⋅100
Mean absolute error
MAE
1
N ∑N
i=1 |yi – ̂yi|
Root mean squared error
RMSE
√
∑Ni=1(yi– ̂yi)2
N
Coefficient of determination
R2
1 – ∑Ni=1(yi– ̂yi)2
∑Ni=1(yi– ̄y)2
https://doi.org/10.1371/journal.pone.0318701.t002
2. Results
2.1. Performance of AVMs
To evaluate the increase in performance due to the addition of different variables, we train
multiple AVMs using the variables presented in Table 1. First, we train a model using only the
BV. Then, we sequentially add the rest of the variables. The idea is to add as many variables
of each type as possible until there is no significant improvement in the model performance,
discarding the characteristics that do not improve at least 3 of the 5 performance metrics
presented in Table 2, compared to the previous model in the sequence.
For each model’s performance evaluation, we used 5-fold cross-validation, implying that
five iterations were made, and in each one, 80% of the dataset was used as training and the
remaining 20% as validation.
Table 3 summarizes the validation results of each model of the sequence. We only present
the variables that increased the previous model’s performance in at least 3 of the 5 perfor-
mance metrics presented in Table 2, see S3 Table for more details. The column Variables
describes the variables used to train each model. In this case, some similar variables were
grouped to ease the interpretation. For example, the group “amenities” contains the variables
parking_lots, swimpool, terrace and storage_units, while the group “distances” contains the
distances to services that improved the performance of the model.
Table 4 shows that the AVM that presents the best performance contains the seven base
variables, five structural variables obtained from the property description, 10 location vari-
ables, two neighborhood variables, and 10 image variables linked to the photos of property
features (bathroom, bedroom, living room, dining room, interior, exterior, balcony, laundry
room, floor plans and swimming pool), for a total of 34 predictor variables.
Table 3. Summary of the performances of AVMs in 5-fold cross-validation with different sets of variables.
Models
Variables
MAPE
MdAPE
MAE
RMSE
R2
Base Model
Base Variables (BV)
11.59%
7.87%
906.66
1513.44
0.9412
Model 1
Base Model +
quality_municipality
11.46%
7.81%
898.40
1501.93
0.9421
Model 2
Model 1 + amenities
10.94%
7.46%
863.40
1448.80
0.9461
Model 3
Model 2 + new_house
10.88%
7.44%
860.92
1446.02
0.9463
Model 4
Model 3 + simce_4b
10.72%
7.30%
849.60
1430.32
0.9475
Model 5
Model 4 +
nearby_schools_3km
10.63%
7.23%
844.36
1424.25
0.9479
Model 6
Model 5 + 9 distances
10.24%
6.88%
816.99
1391.43
0.9503
Final Model
Model 6 + 10 features
10.07%
6.84%
807.53
1373.23
0.9516
https://doi.org/10.1371/journal.pone.0318701.t003
PLOS ONE https://doi.org/10.1371/journal.pone.0318701
March 25, 2025
11/ 27


===== PAGE 12 (score 30, numeric-paren 0, table-words 4, var-words 22) =====

PLOS ONE
A comparative analysis of automated valuation models: Machine learning and spatially-adjusted hedonic pricing
Table 4. Independent variables used in the final model.
Final Model
Dependent variable: ln(price)
Independent variables used:
1. bathrooms
2. bedrooms
3. latitude
4. longitude
5. building_type
6. seller_type
7. total_surface
8. quality_municipality
9. parking_lots
10. swimpool
11. terrace
12. storage_units
13. new_house
14. simce_4b
15. nearby_schools_3km
16. dist_subway
17. dist_park
18. dist_hospital
19. dist_police
20. dist_university
21. dist_mall
22. dist_penitentiary
23. dist_industry
24. dist_dump
25. photo_bathroom
26. photo_bedroom
27. photo_livingroom
28. photo_diningroom
29. photo_interior
30. photo_exterior
31. photo_balcony
32. photo_laundryroom
33. photo_floorplans
34. photo_swimpool
Unused independent variables:
1. simce_8b
2. simce_2m
3. nearby_schools_1km
4. nearby_schools_2km
5. dist_school
6. photo_kitchen
7. photo_patio
8. photo_gym
9. photo_gameroom
10. photo_street
11. photo_parkinglot
https://doi.org/10.1371/journal.pone.0318701.t004
As Table 3 shows, there was a significant impact on the performance metrics when the
variables obtained from the description were used, since they determine the extra value of
properties with parking lots, terraces, swimming pools, and storage units, which are ameni-
ties that undoubtedly affect the price of a property. Also, using the variable new_house helped
to differentiate the used properties from the new ones, which is a relevant aspect when selling
a property. Another set of variables that also significantly reduced the error when estimating
the value of the properties were the distances to services, since they provided details related to
the added value of the property, and the externalities of unwanted services, and it also allowed
to differentiate properties within the same municipality. Finally, the image variables reduced
the error even when the model already incorporated a wide range of information about the
dwelling and its surroundings. This can be explained since these variables provided details
that were not contained in the property description, such as the presence of balconies and
laundry rooms. These variables also allowed prices to be differentiated when the properties
did not include critical photos such as those of bathrooms, living rooms, dining rooms, and
bedrooms, which may be related to the omission of information given the quality and visual
appeal of these settings.
Regarding the variables that were not used in the final model, the simce_8b and simce_2m
attributes were discarded because while they improved the model’s performance, they had
a lower impact than simce_4b. The same occurred with the variables related to the number
of nearby schools. The variables calculated within a one and two kilometer radius did not
enter the final model because the variable with a three-kilometer radius achieved a greater
improvement in performance. On the other hand, the image variables that did not pro-
vide relevant information to the model were the kitchen, patio, game room, gym, street,
and parking lot features, which failed to improve the model’s performance in at least three
metrics.
2.1.1. Relative importance of variables To use the TreeSHAP algorithm, a model was fit-
ted with simple validation. Therefore, the dataset was randomly divided, using 80% to fit the
PLOS ONE https://doi.org/10.1371/journal.pone.0318701
March 25, 2025
12/ 27


===== PAGE 20 (score 54, numeric-paren 0, table-words 11, var-words 18) =====

PLOS ONE
A comparative analysis of automated valuation models: Machine learning and spatially-adjusted hedonic pricing
As expected, Table 7 shows that the SAR model handles the spatial dependence better
than the traditional OLS model by reporting a Moran’s I close to 0. The non-standardized
coefficients of the OLS and SAR models are presented in Table 8.
From Table 8, it can be seen that the number of bathrooms, number of bedrooms, and the
quality of the municipality had overestimated coefficients in the OLS model, which was cor-
rected in the SAR model. On the other hand, some distance variables drastically changed their
behavior, such as the distance to a park, which changed from having a statistically significant
negative coefficient to a statistically significant positive one. In contrast, the distance to an
industry stopped to have a statistically significant impact in the SAR model. In addition, the
Table 7. Moran’s I for the residuals of the OLS and SAR models.
Model
Statistic I
OLS
0.1337
SAR
0.0751
https://doi.org/10.1371/journal.pone.0318701.t007
Table 8. Non-standardized coefficients of the OLS and SAR models (∗∗p-value<0.01.)
OLS (R2 = 0.8654)
SAR (R2 = 0.8774)
Coefficient
Std Error
Coefficient
Std Error
constant
3.7782∗∗
0.0416
1.9507∗∗
0.0515
bathrooms
0.1798∗∗
0.0024
0.1712∗∗
0.0023
bedrooms
0.0814∗∗
0.0021
0.0752∗∗
0.0021
building_type
0.0874∗∗
0.0057
0.0957∗∗
0.0055
seller_type
-0.0662∗∗
0.0044
-0.0649∗∗
0.0043
total_surface
0.0008∗∗
0.0000
0.0007∗∗
0.0000
quality_municipality
0.0273∗∗
0.0003
0.0178∗∗
0.0003
parking_lots
0.0159∗∗
0.0013
0.0139∗∗
0.0012
swimpool
0.0002
0.0034
0.0021
0.0033
terrace
0.0582∗∗
0.0032
0.0465∗∗
0.0031
storage_units
0.0304∗∗
0.0016
0.0289∗∗
0.0015
new_house
0.0753∗∗
0.0073
0.0668∗∗
0.0070
simce_4b
0.0082∗∗
0.0001
0.0036∗∗
0.0001
nearby_schools_3km
-0.0008∗∗
0.0000
0.0014∗∗
0.0000
dist_subway
0.0045∗∗
0.0011
0.0054∗∗
0.0010
dist_park
-0.0019
0.0017
0.0087∗∗
0.0016
dist_hospital
0.0038∗∗
0.0010
-0.0084∗∗
0.0009
dist_police
0.0263∗∗
0.0020
0.0194∗∗
0.0019
dist_university
-0.0176∗∗
0.0013
-0.0121∗∗
0.0011
dist_mall
-0.0039∗∗
0.0014
0.0052∗∗
0.0013
dist_penitentiary
-0.0050∗∗
0.0006
-0.0035∗∗
0.0006
dist_industry
0.0096∗∗
0.0009
-0.0006
0.0009
dist_dump
0.0019∗∗
0.0005
-0.0045∗∗
0.0005
photo_bathroom
-0.0506∗∗
0.0035
-0.0450∗∗
0.0034
photo_bedroom
-0.0015
0.0038
-0.0015
0.0037
photo_livingroom
0.0492∗∗
0.0039
0.0438∗∗
0.0037
photo_diningroom
0.0488∗∗
0.0032
0.0441∗∗
0.0030
photo_interior
0.0017
0.0030
0.0036
0.0029
photo_exterior
-0.0011
0.0031
-0.0004
0.0030
photo_balcony
0.0249∗∗
0.0030
0.0209∗∗
0.0029
photo_laundryroom
-0.0326∗∗
0.0039
-0.0249∗∗
0.0037
photo_floorplans
0.0445∗∗
0.0054
0.0422∗∗
0.0052
photo_swimpool
-0.0092∗∗
0.0034
-0.0115∗∗
0.0032
W_ln_price
-
-
0.4271∗∗
0.0080
https://doi.org/10.1371/journal.pone.0318701.t008
PLOS ONE https://doi.org/10.1371/journal.pone.0318701
March 25, 2025
20/ 27


===== PAGE 21 (score 23, numeric-paren 0, table-words 7, var-words 8) =====

PLOS ONE
A comparative analysis of automated valuation models: Machine learning and spatially-adjusted hedonic pricing
distance to a hospital and the distance to a dump changed from positive to negative param-
eters. The opposite occurred with the distance to a mall, which underwent a coefficient sign
change from negative to positive. Regarding the neighborhood variables, the average SIMCE
score of elementary schools reduced the positive impact on the price in the SAR model. In
contrast, the number of nearby schools within a 3-kilometer radius went from having a neg-
ative coefficient to a positive one, maintaining a statistically significant impact on the price.
Note that this effect is the opposite to the one obtained with the IML approach. In other
words, the insight obtained from the IML approach, regarding the effect of schools, only coin-
cides with that of the OLS model, i.e., the model with no spatial corrections. Moreover, note
that the spatial autoregressive parameter (𝜌) had a value of 0.4271, confirming a positive
spatial dependence on the price of residences, thus suggesting a biased result from the IML
approach. Additionally, although the OLS and SAR models were fitted primarily for quali-
tative comparisons with the AVM approach, we report their performance results in Table 9.
As expected, both models show significantly worse prediction performance than the AVMs
(Table 3), a result that is consistent with previous literature [21,44].
Finally, we compared the relative importance that the three models assigned to the vari-
ables. To do this, we standardized the attributes of the OLS and SAR models, so their unit of
measurement was the number of standard deviations. The standardization formula or Z-score
is shown in Eq 8.
Z = x – 𝜇
𝜎
(8)
In Eq 8, x is the variable’s value, 𝜇is the mean, and 𝜎is the standard deviation. Perform-
ing this transformation allows the coefficients of the models to be standardized, so the abso-
lute value of the coefficients can be analyzed to see which characteristics have a more signif-
icant effect on the price of homes. On the other hand, the relative importance of the Light-
GBM model variables is assigned by the SHAP algorithm. For comparative purposes, we
excluded the latitude and longitude variables from the ranking since they were used in the
spatial weights matrix in the SAR model.
Table 10 shows a close relationship between the importance assigned to the variables in
each model, even though the methods and approaches were different. In the five most impor-
tant variables of each model, four attributes stand out: total surface area, quality of the munic-
ipality, average SIMCE score of elementary schools, and the number of bathrooms. Although
the rankings assigned for these attributes are not the same, the three models agree that these
variables are the ones that have the most significant impact on the price. Due to the spa-
tial correction made by the SAR model, the quality of the municipality decreased its effect
on the price, causing the number of bathrooms to be the most important variable. A simi-
lar change occurred with the average SIMCE score of elementary schools, which reduced its
impact on property values, relegating it to fifth place and moving the number of bathrooms
to fourth place in the ranking. On the other hand, the distance variables lost importance
in econometric models because their interactions with the prices are not linear, which is a
characteristic that the LightGBM model can capture.
Table 9. Performance summary of OLS and SAR models in 5-fold cross-validation.
Model
MAPE
MdAPE
MAE
RMSE
R2
OLS
21.13%
16.86%
1887.85
3356.50
0.7103
SAR
20.92%
17.13%
1893.09
3115.85
0.7405
https://doi.org/10.1371/journal.pone.0318701.t009
PLOS ONE https://doi.org/10.1371/journal.pone.0318701
March 25, 2025
21/ 27