# RECORD 1076844f99e9f35c7768a4e64f2c167ab26b5afd
# TITLE: Predicting Housing Prices in Riyadh: A Comparative Analysis on Machine Learning and Hedonic Regression
# YEAR: 2025.0  VENUE: Innovation in Statistics and Probability  DOI: 10.64389/isp.2025.01220
# AUTHORS: Mohamed Ezzeldin; Mahmoud Abouagwa
# PAGES: 19  CANDIDATE TABLE PAGES (1-based): [1, 5, 6, 8, 9, 11, 12, 13, 15]
# ABSTRACT: This study investigates the performance of hedonic regression against artificial neural networks and random forest for real estate price prediction in Riyadh, Saudi Arabia. Using data from a major property platform collected in Q3 2024, the analysis spans 2,997 residential units across five major districts. The study is situated within the context of Vision 2030 and addresses the challenges of real estate valuations in a rapidly evolving urban environment. Twelve locational and structural features were analyzed using a 70/30 training-testing split. Random Forest led with test R²=0.9065, RMSE=0.1911, MAE=0.1353, outperforming Artificial Neural Network (R²=0.8968, RMSE=0.2008, MAE=0.1450) and Hedonic Regression (R²=0.8034, RMSE=0.2673, MAE=0.1970). Random Forest feature importance: North of Riyadh (0.768659), West of Riyadh (0.612986), building type (0.180799), entrances (0.171134), size (0.127103). Random Forest and Artificial Neural Network excelled in capturing non-linear and spatial price trends in Riyadh’s housing market. This study supports the incorporation of machine learning models into workflows for predicting property prices, especially in cities like Riyadh that are quickly urbanizing. The research provides a useful foundation for improving real estate valuation procedures in accordance with Saudi Arabia's Vision 2030 goals by demonstrating the predictive benefits of RF over conventional methods.


===== PAGE 1 (score 13, numeric-paren 5, table-words 0, var-words 7) =====

Innovation in Statistics and Probability 
1(2), 26–44 
DOI:10.64389/isp.2025.01220 
https://sphinxsp.org/journal/index.php/isp/ 
 
 
 
 
 
Research article 
Predicting Housing Prices in Riyadh: A Comparative Analysis on Machine Learning 
and Hedonic Regression 
Mohamed R. Ezzeldin11,* and Mahmoud Abouagwa 2,3 
1 Deanship of Educational Services, Qassim University, Qassim 51452, Saudi Arabia; m.ezzeldin@qu.edu.sa 
2 College of Business, City University Ajman, Ajman 18484, United Arab Emirates, m.abouagwa@cu.ac.ae 
3 Department of Mathematical Statistics, Faculty of Graduate Studies for Statistical Research, Cairo University, Giza 
12613, Egypt; mahmoud.aboagwa@cu.edu.eg  
 
*  Correspondence: m.ezzeldin@qu.edu.sa 
 
ARTICLE INFO 
Keywords: 
Real estate valuation, Riyadh housing 
market, Hedonic regression, Random 
Forest, Neural networks, Machine 
learning in real estate.  
Mathematics Subject Classification: 
62G08, 68T07 
Important Dates: 
Received: 14 August 2025  
Revised: 21 October 2025  
Accepted: 25 November 2025  
Online: 26 November 2025 
 
 
Copyright © 2025 by the authors. Published 
under Creative Commons Attribution (CC BY) 
license. 
ABSTRACT 
This study investigates the performance of hedonic regression against 
artificial neural networks and random forest for real estate price prediction in 
Riyadh, Saudi Arabia. Using data from a major property platform collected in 
Q3 2024, the analysis spans 2,997 residential units across five major districts. 
The study is situated within the context of Vision 2030 and addresses the 
challenges of real estate valuations in a rapidly evolving urban environment. 
Twelve locational and structural features were analyzed using a 70/30 training-
testing split. Random Forest led with test R²=0.9065, RMSE=0.1911, 
MAE=0.1353, outperforming Artificial Neural Network (R²=0.8968, 
RMSE=0.2008, MAE=0.1450) and Hedonic Regression (R²=0.8034, 
RMSE=0.2673, MAE=0.1970). Random Forest feature importance: North of 
Riyadh (0.768659), West of Riyadh (0.612986), building type (0.180799), 
entrances (0.171134), size (0.127103). Random Forest and Artificial Neural 
Network excelled in capturing non-linear and spatial price trends in Riyadh’s 
housing market. This study supports the incorporation of machine learning 
models into workflows for predicting property prices, especially in cities like 
Riyadh that are quickly urbanizing. The research provides a useful foundation 
for improving real estate valuation procedures in accordance with Saudi 
Arabia's Vision 2030 goals by demonstrating the predictive benefits of RF over 
conventional methods.
 
 
 
1. Introduction 
Housing constitutes one of the primary human needs, and a substantial portion of household income is typically 
allocated to its purchase. As such, residential real estate is a critical asset class in household portfolios, and fluctuations 
in housing prices can have substantial macroeconomic repercussions [1,2]. Urban congestion and uneven development 


===== PAGE 2 (score 19, numeric-paren 0, table-words 1, var-words 17) =====

27 
 
 
 
 
 
contribute to regional inflation in property markets, making housing price forecasting a valuable tool for planners and 
investors. This study investigates real estate price prediction in Riyadh, Saudi Arabia, within the context of urban 
growth, market complexity, and data-driven housing policy under Vision 2030. Saudi Arabia’s urban housing market 
is undergoing rapid transformation, driven in part by strategic initiatives encapsulated in Vision 2030 the Kingdom’s 
ambitious plan to diversify its economy and enhance quality of life. Vision 2030’s housing related goals include 
expanding affordable housing supply, promoting private sector engagement, and improving urban infrastructure and 
governance. These reforms have crucial effects on market dynamics and housing prices, making accurate predictive 
modeling essential for policy and investment decisions. 
In Riyadh, the capital of Saudi Arabia, rapid population growth, extensive infrastructure development, and shifting 
housing preferences have resulted in a highly segmented real estate market. The Kingdom's Vision 2030 initiative has 
further intensified housing market dynamics by encouraging private sector participation, modern urban planning, and 
enhanced home ownership [3]. Understanding price formation in such a rapidly evolving environment requires models 
that can handle nonlinearity, spatial diversity, and heterogeneous consumer preferences. The hedonic pricing method 
is a commonly used approach to estimate property values by decomposing price into the implicit value of its 
characteristics [4,5]. These characteristics are typically grouped into structural (e.g., size, age), locational (e.g., 
proximity to metro stations, CBDs), and environmental (e.g., views, air quality) factors. However, one challenge with 
hedonic regression lies in its reliance on linear or semi-log functional forms and pre-specified relationships between 
variables and price [2,6]. In practice, the relationship between housing features and value is often nonlinear and context 
dependent, particularly in complex urban markets like Riyadh. Several efforts have been made to address this limitation 
using advanced statistical techniques such as Box-Cox transformations and geographically weighted regression or 
semi-parametric models [7,8]. More recently, machine learning methods particularly artificial neural networks (ANN) 
and Random Forest (RF) have been deployed in real estate modeling for their ability to handle high dimensional 
nonlinear relationships without strict assumptions on functional form [9,10,11]. To address these challenges, this study 
compares conventional hedonic regression models against modern machine learning approaches Artificial Neural 
Networks (ANN) and Random Forests (RF) for price prediction of residential properties in Riyadh. Using a dataset of 
2,997 residential units collected in Q3 2024 from a widely used real estate platform (Aqar), we apply cross-validation 
to optimize model parameters and assess performance using metrics such as Root Mean Square Error (RMSE), Mean 
Absolute Error (MAE), and R-squared. By focusing on a localized dataset and comparing traditional and machine 
learning techniques. This study aims to provide actionable insights for urban economists, housing policymakers, and 
private sector developers in Saudi Arabia. The study aims to compare the predictive performance of hedonic regression, 
ANN, and RF for housing price prediction in Riyadh, providing insights for urban planners and policymakers under 
Vision 2030. The study was conducted using data collected from Riyadh, Saudi Arabia, in the third quarter of 2024. 
1.1. Housing development and data issues 
Cities have quite different housing markets in terms of pricing and geography. For instance, studies conducted in 
Western cities have demonstrated a pronounced pattern of segmentation brought about by economic and racial 
exclusion [12]. Apart from the ethical and political implications of these developments, a single property's value is 
constantly influenced by environmental, structural, and locational factors in its surrounding community [4]. While the 
adage "location, location, location" is crucial for assessing investment possibilities in the housing market [13], 
environmental and locational factors (such as the distance to the central business district [14] are sometimes more 
challenging to measure.  
These foundational insights demonstrate the multifaceted influences shaping housing markets and underscore the 
critical role of location and urban amenities in determining property prices worldwide. However, contrasting dynamics 
emerge when we examine markets in different cultural and regional contexts. 
Riyadh and other Saudi Arabian cities have their own sets of housing markets, more so depending on the area 
within that city and price range; the differences in these markets emanate from the peculiar cultural, economic, and 
social dynamics of the Kingdom. While Western cities may have housing market segmentation based on racial 
segregation and economic exclusion, in Saudi Arabia, the housing market characteristics are heavily based on the 
proximity to religious, commercial, and urban amenities. For example, the value of the property is highly affected by 
its closeness to areas such as mosques, shopping centers, and planned transport facilities.  
Such localized factors, unique to Riyadh, influence spatial urban patterns and real estate valuation distinctly, 
highlighting the need for models sensitive to these socio-cultural and infrastructural determinants. 
In Saudi Arabia, since the introduction of Vision 2030, several governmental reforms have been undertaken to 
tackle the housing problems, and a transition from a government allocated housing system to market forces has been 
provided. Given the Sakani programs, the opportunities for home ownership have opened for its citizens, thus changing 
the dimensions of urban housing. Housing in Saudi Arabia was once characterized by traditional neighborhoods where 
families lived close to essential services, like schools, retail outlets, and healthcare facilities. Development dependent 


===== PAGE 5 (score 25, numeric-paren 4, table-words 3, var-words 15) =====

30 
 
 
 
 
 
price model, every feature of nonhomogeneous items offers a distinct profit or level of usefulness. The fair and realistic 
pricing of an item in its market is often ascertained using this approach. This strategy is based on the consumer theory 
that Lancaster [15] established, and Rosen [5] expanded to the real estate industry. The multiple regression model and 
the hedonic method serve the same functions. Regression analysis may make use of the hedonic idea. The price of a 
property is the dependent variable, while its qualities are the independent variable. In real estate and related markets, 
including the valuation of any item, regression analysis is referred to as the hedonic pricing model [20]. The hedonic 
analysis focuses on the relationship between a product's attributes and its price. Residential properties are rented or 
purchased because they provide utility or satisfaction from the amenities they have. Each home, whether it is an 
apartment, duplex, or house, has its features that affect its desirability. Structural attributes (land size, interior square 
footage, age of the building, layout of rooms, and additional facilities such as garages or air conditioning) are defined 
as structural. Locational aspects such as proximity to employment centers, retail stores, and recreational venues affect 
property values intrinsically. Environment and sociodemographic factors impact values because, through these two, 
the desirability of a neighborhood is defined. Similar forces affect how willing people are to pay for their housing unit. 
The most basic goal of hedonic analysis within real estate is essentially to quantify the relationship of structural and 
locational attributes to property prices. Observable, measurable items number of bedrooms, distance to school is 
certainly there, but not every factor observable through the lens is directly linked with price determination in the market. 
Through dynamics of supply and demand, the prices in the housing market act in their unique way, whereby their action 
on each unit depends upon a particular type of attribute and a more general market behavior, and therefore, the price 
of a housing unit is unique. Hedonic analysis aims to estimate the hedonic function, according to which each input 
attribute would set the market price. The hedonic analysis assumes that each characteristic of a property has its market, 
conditional on its supply and demand. Thus, every characteristic has its "hedonic price," which informs how much it 
contributes value to the property. Understanding this function is significant for three important reasons: 
Property valuation: It facilitates the determination of the market price of a housing unit based on the attributes of 
that unit, thus contributing significantly to decisions in real estate appraisals and investment decisions. 
Housing price indexing: Cost of living comparisons between regions are possible through price indices of constant 
quality. For instance, comparisons of housing prices in Riyadh and Jeddah will demand adjusting for differences across 
properties. 
Policy analysis: Hedonic models provide a means to value certain implicit benefits from specific attributes, such 
as air quality, which improves insight into the social costs and benefits of environmental and infrastructural policies. 
By using hedonic analysis, the construction of house price indices, which allow analysis for changes in market 
conditions while removing effects due to particular property attributes, is possible. In this study, hedonic regression is 
applied to estimate the implicit prices of housing attributes in Riyadh, identifying how structural and locational factors 
influence property values. To conduct the analysis, categorical variables were transformed into numerical variables, 
including Building Type (e.g., apartment, villa) and Location (North, South, West, East, Central Riyadh). In the hedonic 
regression analysis, dummy variable coding was employed, wherein East Riyadh served as the reference category for 
Location, while for Building Type, villas were set as the reference category. All categorical variables were one-hot 
encoded for binary input to the ANN, thereby ensuring compatibility with the input layer of the neural network. A 
product is sold as a collection of characteristics, which may be expressed in a matrix f(x1, x2, x3, … , xn) [34]. This is 
the basic tenet of the hedonic pricing model. As a result, the product's market price may be represented as a function 
of these attributes, or 𝑦=  𝑓(𝑥1, 𝑥2, 𝑥3, … , 𝑥𝑛). Equation (1) illustrates the hedonic pricing model's generic form: 
y =  α +  β𝐗 +  e,  
(2.1) 
where y is the price vector, e is the random error vector, α, β are the coefficient vectors, and X is the matrix of 
independent variables. Scholars have observed that heteroscedasticity can arise from a skewed distribution of both Y 
and xn [30, 33]. Therefore, as the model was being implemented, we used the semi-log model.: 
𝑉𝐴𝐿𝑈𝐸 = 𝑒𝑥𝛽𝜀.    
(2.2) 
Thus 
 𝑙𝑛𝑉𝐴𝐿𝑈𝐸= 𝑥𝛽+ 𝜀.    
(2.3) 
 
In this case, the log of an item’s expected price is the sum of its characteristics X times 𝛃, and the marginal price 
of each attribute x is: 
𝑃𝑅𝐼𝐶𝐸 (𝒙) = 𝑒𝑥𝑏.   
(2.4) 
 
Where x is the current level of the characteristic and b is the regression coefficient. The marginal effect of an 
attribute in a semi-log model can be approximated as (𝑒𝑏−1 ) × 100%, representing the percentage change in price 
associated with a one-unit change in the characteristic. Notice that the semi-log form implies that the price of a given 


===== PAGE 6 (score 12, numeric-paren 3, table-words 2, var-words 5) =====

31 
 
 
 
 
 
characteristic varies with its level, i.e., the prices are nonlinear. Compared to its linear version, the semi-log structural 
hedonic pricing model offers several benefits. It allows the value of one feature (like the number of bathrooms) to 
change proportionally with the value of other qualities (like the number of beds), which is its main advantage. A second 
bathroom offers the same value to a one-bedroom home as it does to a five-bedroom home, but this is not the case with 
a linear model. Also, given the skewed nature of housing prices, we apply a semi-log transformation to stabilize 
variance and improve interpretability.  
2.2.ANN methodology 
The computational and nonlinear statistical modeling technique known as ANN is based on human biological 
neurons [45,46]. Aerospace, automotive, banking, defense, electronics, entertainment, finance, insurance, 
manufacturing, medical, oil & gas, robots, speech, securities, telecommunications, and transportation are just a few of 
the industries that have made extensive use of it. In the context of regression, ANN may also produce precise 
predictions. One may think of it as a nonlinear regression technique [5]. Numerous sub-components, including weights, 
nodes, layers, and activation functions, make up a neural network structure. The input layer, which contains the values 
of the independent variables, the hidden layer, which has a specific number of processing units and nodes, and the 
output layer, which provides the estimated values of the dependent variable, are the three primary layers. Typically, 
weights are selected at random from a distribution of values within a predetermined range, such as [-1,1]. Utilizing 
total net, which is defined as the sum of weighted inputs and bias values, the last component, the activation function, 
generates values. There are connections between each of these elements. Similar to traditional regression models, it 
uses the same independent variables as inputs and the dependent variable as an output. 
 
Figure 1: A sample structure of a multilayer neural network 
To minimize the discrepancy between estimated and real dependent values, ANN essentially learns by analyzing 
the dataset itself and updating weights. Figure 1 illustrates a multilayer perception with one hidden layer, depicting the 
connections between input, hidden, and output layers. The input-hidden layer connection may be stated as follows and 
represented in linear form: 
ht(X) = g (W0 + ∑Xi Wit
P
i=1
), 
(2.5) 
 
where g(.) is the sigmoid activation function, 
𝑔 ( 𝑧 ) =
1
1 + 𝑒−𝑧. 
(2.6) 
The weight by 𝑤𝑖𝑡, and the bias value between each variable and the associated hidden node (t) by W0. This value, 
ℎ𝑡(𝑋), is only the result of the hidden node t. The result value may be similarly described as a linear combination of 
these nodes after the number of hidden nodes in the hidden layer has been determined [46]: 
f(X) = η0 + ∑ηt ht
M
t=1
. 
(2.7) 
In this case, the estimated result values are represented by 𝑓(𝑥). The sum of the squared residuals is minimized or 
reduced by updating the parameters of an ANN model. Several learning algorithms, including the popular 
backpropagation method put out by Kim and Nelson, are employed to carry out this updating process. Note that 
reaching the global optimal solution is not guaranteed [47]. In this study, ANN is used to complement hedonic 
regression by capturing nonlinear relationships and interactions between housing attributes in Riyadh. The ANN model 
employs a multilayer perceptron with one or two hidden layers, where the number of nodes per layer was optimized 


===== PAGE 7 (score 16, numeric-paren 1, table-words 2, var-words 10) =====

32 
 
 
 
 
 
through 5-fold cross-validation. A sigmoid activation function is applied, and the model is trained using 
backpropagation with a learning rate of 0.01 for 500 epochs. Regularization techniques, including L2 regularization (λ 
= 0.01) and 20% dropout, were applied to the hidden layers to mitigate overfitting. Early stopping was implemented 
after 10 epochs without improvement in validation loss. Optimization of ANN architecture was carried out using 5 
folds of cross-validation applied to a grid of one or two hidden layers with node counts {5, 10, 15, 20}. The smallest 
RMSE (0.2581) on the validation set was obtained with one hidden layer with 5 nodes and the second with 15 nodes. 
Alternative deeper architectures (ex., 3 layers) increased the complexity in computations (training time escalated from 
2.3 to 4.1 min) but did little to enhance accuracy (RMSE=0.2592). To minimize the overfitting issue, L2 regularization 
(λ = 0.01) and 20% dropout rates have been used on hidden layers. From sensitivity tests with λ = {0.001, 0.1}, the 
decrease in RMSE was minimal (<0.002), while increasing dropout to 30% raised test RMSE to 0.2051. Hence, 20% 
became the optimal dropout rate. Early stoppage occurred after 10 epochs, where validation loss ceased to improve. 
This brought a gap in train and test R² that were previously 0.015 apart to just 0.0074. 
2.3.RF methodology 
A Random Forest is defined formally as follows: it is a classifier consisting of a collection of tree structured 
classifiers {hk(x, Tk)}, k =  1, 2, . . . , L, Where Tk are independent identically distributed random samples, and each tree 
casts a unit vote for the most popular class at input x. As mentioned, RF employs the same method of bagging to 
produce random samples of training sets (bootstrap samples) for each random tree. Each new training set is built, with 
replacement, from the original training set. Thus, the tree is built using the new subset and a random attribute selection.  
The best split on the random attributes selected is used to split the node. The trees grown are not pruned. The use of 
the bagging method is justified for two reasons [47]: the use of bagging seems to enhance performance when random 
attributes are used; and bagging can be used to give ongoing estimates of the generalization error of the combined 
ensemble of trees, as well as estimates for strength and correlation. These estimates are performed out-of-bag. In a 
Random Forest, the out-of-bag method works as follows: given a specific training set T, generate bootstrap training 
sets Tk, construct classifiers {hk(x, Tk)} and let them vote to create the bagged classifier. For each (x, y) in the training 
set, aggregate the votes only over those classifiers for which Tk does not contain (x, y).  This is the out-of-bag classifier. 
Then the out-of-bag estimate for the generalization error is the error rate of the out-of-bag classifier on the training set. 
The error of a forest depends on the strength of the individual trees in the forest and the correlation between any two 
trees in the forest. Strength can be interpreted as a measure of performance for each tree. Increasing the correlation 
increases the forest error rate, and increasing the strength of the individual trees decreases the forest error rate since a 
tree with a low error rate is a strong classifier. Reducing the number of random attributes selected reduces both the 
correlation and the strength [48]. The RF model used 500 trees having max depth=10, min samples split=5, min samples 
leaf=3 selected using the grid search over {100, 300, 500} trees and depth {5, 10, 15}. This setup was a tradeoff sweet 
in terms of accuracy (test RMSE = 0.1911) and overfitting (test-train RMSE gap = 0.0334 vs. 0.0451 for depth=15). 
ANN and RF techniques are selected among other models of Machine Learning for their capability of forecasting the 
nonlinear relationships and complex interactions due to the Vision 2030 project housing factors, which would include 
behavioral factors that influence the housing market of Riyadh. Unlike linear methods such as Lasso Regression, or 
simpler machine learning methods such as k-NN, ANN constructs feature hierarchies through the neural network 
multilayer perceptron and the RF, which is composed of 500 decision trees that smoothen the nonlinear thresholds 
using heterogeneous records (e.g., Size: 56-930 m²). These two have good robustness against mixed attribute datasets, 
but not SVM, which has multiple kernel tuning requirements. Their established effectiveness in real estate [11] and 
balance of predictive power (RF: test R² = 0.9065) with interpretability (RF feature importance) align with the study's 
benchmark goal using hedonic regression against Machine Learning (ML). Also, do not require much of the resource 
intensive alternatives (for example, deep learning), making them a perfect fit for this analysis. 
2.4.Case study and data 
The very highly renowned Riyadh, the capital city of Saudi Arabia, has long been recognized as the cultural, 
financial, and political hub of the country. The capital city has a population greater than 7 million people and is 
surrounded by the King Khalid Road, which outlines an urban area of about 1,500 square kilometers. Areas beyond 
this road were not included in the research as they are largely peripheral with limited support for urban functions. The 
main data has been retrieved from a well-known real estate website, Aqar (https://sa.aqar.fm/), during the third quarter 
of 2024 and georeferenced by property addresses. We collected 2,997 units and 12 variables, which are given with 
some descriptive statistics in Table 1. The dependent variable is listing price rather than transaction price, as actual 
transaction data are not publicly available in Saudi Arabia. Listing prices reflect seller expectations but remain highly 
correlated with transaction prices in this market, as noted in prior empirical work. The dataset belongs to five districts, 
including the North, South, West, East, and Center of Riyadh. The variables used are Building Type, Bedroom, Living 
Room, Bathroom, Kitchen, Size, Street Width, Floor, Age, Entrance Number, Elevator, and Location. The 2,997 units 


===== PAGE 8 (score 36, numeric-paren 0, table-words 9, var-words 16) =====

33 
 
 
 
 
 
represent all available residential listings within Riyadh’s urban boundaries during the data collection period, with only 
properties having complete data for all 12 variables included. The variable “Location” is operationalized as a dummy 
variable for each of the five districts to account for spatial price variations. Key variables include Building Type 
(categorical, e.g., apartment, villa), Size (continuous, square meters), and Age (continuous, years), each expected to 
influence property prices based on their contribution to utility or desirability. 
While actual transaction prices are preferable, they remain inaccessible in Saudi Arabia due to privacy regulations 
and limited public disclosure requirements. We rely on listing prices, which represent seller expectations and are 
publicly available through the Aqar platform. Research from comparable emerging markets suggests listing prices are 
highly correlated with transaction prices, particularly in regulated markets with standardized listing practices. In Saudi 
Arabia's context, the Aqar platform is the most widely used real estate marketplace, and listings undergo verification 
processes that reduce the likelihood of inflated expectations. Moreover, Vision 2030 initiatives have introduced greater 
transparency in the housing sector, narrowing the gap between listing and final transaction prices. 
 
3. Results and findings 
3.1.Data set and preprocessing 
The issue of multicollinearity is significant due to the reference to near-linear dependences among 
independent variables. According to Montgomery [49], the presence of multicollinearity can result in unstable 
regression coefficients with high variances/covariances and absolute values. The condition index and variance 
inflation factor are used to assess the presence or absence of multicollinearity. The thresholds are set at 10 for 
the Variance Inflation Factor (VIF). These criteria show that there is no multicollinearity between independent 
variables. The outcome is shown in the hedonic result table. The dataset has been split into train and test data. 
Between them, with 70% for training and 30% for testing, respectively. Training data was used to fit the models, 
while testing data was used to evaluate them. RMSE, R-squared, and MAE have been computed and provided 
as performance metrics. The outcomes of hedonic regression analysis and artificial neural networks have been 
analyzed using Python 3.12 software. Which are presented in Table 1. 
Table 1: Descriptive Statistics of Property Characteristics. 
Variable 
Min 
Max 
Mean 
Std. Deviation 
Bedrooms 
1 
26 
4.19 
1.78 
Living Rooms 
1 
6 
1.50 
0.82 
Bathrooms 
1 
8 
3.32 
1.11 
Kitchens 
1 
6 
1.04 
0.29 
Size (m²) 
56 
930 
199.44 
114.05 
Floor Number 
0 
10 
1.57 
0.88 
Age (years) 
0 
20 
2.48 
3.99 
Number of Entrances 
1 
2 
1.02 
0.32 
 
In Table 1, the dataset's descriptive statistics are compiled. The figures show that most of the residences are 
apartments, which are typically smaller than villas and feature fewer bedrooms, living rooms, and bathrooms. 
Typically, apartments are 152.74 square feet in size, with 1–7 bedrooms, 1-3 living spaces, and 1–5 baths. On 
the other hand, villas are more expensive and diverse, with an average size of 347.13 square units, ranging from 
2 to 26 bedrooms, 1-6 living rooms, and 1 to 8 bathrooms. The average size of a property is 199.44 square units, 
with 4.19 bedrooms, 1.50 living rooms, and 3.32 bathrooms, according to the total figures.  
The dataset underwent rigorous quality control during data collection from the Aqar platform. Only 
properties with complete information for all 12 variables were included in the analysis, resulting in the final 
sample of 2,997 residential units. Properties with missing values in any field were excluded during the initial 
data cleaning phase. 
Regarding outliers, descriptive statistics reveal property sizes ranging from 56 to 930 square meters, which 
represent legitimate market variations rather than data errors. The semi-log transformation applied to the 
dependent variable (housing prices) helped stabilize variance and mitigate the influence of extreme values 
without requiring arbitrary outlier removal. Additionally, the Variance Inflation Factor (VIF) analysis confirmed 
no multicollinearity issues (all VIF values below 4), and the Breusch-Pagan test (p = 0.13) confirmed 
homoscedasticity, indicating that the data structure was appropriate for modeling 
 
3.2.Hedonic Regression Results 
The hedonic regression results are given in Table 2. Which includes the hedonic coefficient, standard error, 


===== PAGE 9 (score 35, numeric-paren 1, table-words 5, var-words 12) =====

34 
 
 
 
 
 
t-statistics, P-value, and VIF calculated below. 
 
Table 2: Hedonic regression model result. 
 
B 
Std. Error 
t 
Sig. 
Tolerance 
VIF 
Building Type 
-0.473 
0.022 
-21.673 
0.000 
0.275 
3.642 
Bed Room 
0.012 
0.004 
2.837 
0.005 
0.447 
2.24 
Living Room 
0.045 
0.008 
5.349 
0.000 
0.491 
2.036 
Bathroom 
0.03 
0.006 
5.007 
0.000 
0.533 
1.877 
Kitchen 
-0.005 
0.019 
-0.259 
0.795 
0.751 
1.331 
Size 
0.002 
0 
27.336 
0.000 
0.36 
2.775 
Street Wide 
0.001 
0.001 
1.468 
0.142 
0.821 
1.217 
Floor 
-0.022 
0.006 
-3.599 
0.000 
0.815 
1.228 
Age 
-0.007 
0.001 
-5.051 
0.000 
0.842 
1.188 
Entrance Number 
0.047 
0.016 
2.972 
0.003 
0.942 
1.062 
Elevator 
0.065 
0.011 
5.861 
0.000 
0.796 
1.256 
North of Riyadh 
0.333 
0.013 
24.641 
0.000 
0.653 
1.532 
South of Riyadh 
-0.686 
0.016 
-41.622 
0.000 
0.747 
1.338 
West of Riyadh 
-0.298 
0.013 
-22.715 
0.000 
0.684 
1.462 
Central Riyadh 
-0.169 
0.045 
-3.797 
0.000 
0.938 
1.066 
(Constant) 
14.211 
0.06 
238.054 
0.000 
 
 
F (P-Value) 
840.452 
(0.000) 
 
 
 
 
 
Categorical variables of hedonic regression results indicate a few key trends in the housing market of Riyadh. 
Deciding on physical characteristics and types of buildings is important, with apartments showing very low 
prices compared to villas (-0.473; p < 0.001). The number of bedrooms shows a mild positive effect on house 
prices, with each bedroom adding 1.2% to the price (p = 0.005). Where living spaces are concerned, the effect 
is larger, with each additional living space contributing 4.5% (p < 0.001) to the house price. Bathrooms are also 
positively affecting pricing; an additional bathroom will increase the value by 3% (p <0.001). Interestingly, 
despite its practical significance, the number of kitchens had an insignificant effect on pricing (p = 0.795). This 
may be because of the dataset's low variance (mean = 1.04, SD = 0.29) or Riyadh culture's preference for other 
features, such as location or size, over more kitchens. A more positive coefficient for property size is determined 
at least from the numerical values as one of significance (0.002, p < 0.001) - thus, higher prices are commanded 
for bigger properties. Street width has a positive trend but is statistically insignificant (p=0.142). Higher units 
would be less well priced, as their floor level adversely affects the price (-0.022, p < 0.001). Conversely, older 
units will expect a lower price because of the same negative parameter of the age (-0.007, p < 0.001) of the 
building. The number of entrances influences a positive price valuation on properties (0.047, p = 0.003), with 
the presence of an elevator increasing property values by 6.5% (p < 0.001).  The location is thus a significant 
price driving factor for housing in Riyadh, with substantial distances between neighborhoods and East Riyadh 
(reference category). The properties in North Riyadh command a hefty premium, being priced 33.3% higher than 
East Riyadh (p <0.001). On the other hand, South Riyadh has the most negative effect, with the pricing 68.6% 
lower than the reference area (p < 0.001). Properties in West Riyadh are priced 29.8% lower (p <0.001), while 
Central Riyadh prices are 16.9% lower (p <0.001) in comparison with East Riyadh. Diagnostic tests for the 
model yielded strong results, with VIF values for all low predictor variables being below 4, implying that serious 
multicollinearity can be ruled out, while significance levels for most variables indicate the robustness of the 
model in explaining price variation in the Riyadh housing market. A Breusch-Pagan test (p = 0.13) confirmed 
no significant heteroscedasticity. VIF ranged from 1.062 to 3.642, with a condition index of 8.2, ruling out 
multicollinearity (Montgomery et al., 2012). The regression model demonstrates strong statistical significance, 
with an F-statistic of 840.452 (p < 0.001), indicating that the predictors collectively explain a significant portion 
of the dependent variable's variance. Due to the absence of precise geographic coordinates (latitude and 
longitude) in the dataset, it was not possible to conduct explicit spatial autocorrelation tests such as Moran’s I. 
Therefore, the study incorporates location dummy variables at the district level to partially account for spatial 


===== PAGE 10 (score 26, numeric-paren 4, table-words 10, var-words 1) =====

35 
 
 
 
 
 
dependencies. This limitation suggests caution in interpreting results near district boundaries, where unmodeled 
spatial interactions may persist [22]. This constraint advises using caution when interpreting findings close to 
district boundaries, where unmodeled spatial effects may persist. To partially account for spatial heterogeneity, 
we employed district-level dummy variables (North, South, East, West, and Central Riyadh) in all three models. 
This approach captures broad locational premiums/discounts but has limitations at district boundaries where 
properties may share similar spatial characteristics despite being assigned to different administrative zones. 
 
 
Figure 2: Observed and predicted house prices by hedonic regression 
 
Figure 2 Observed vs. Predicted House Prices by Hedonic Regression. The plot shows a strong linear 
relationship, with predictions closely aligning with actual prices. However, there is a clear underestimation of 
larger price points, suggesting that the model's effective pricing struggles with nonlinear in the Riyadh housing 
market. Strong linearity can be seen in the plot, which indicates a broad capture of price trends by the hedonic 
regression model. However, there is a clear underestimation of larger price points, suggesting that the model's 
effective pricing is nonlinear in the Riyadh housing market. 
 
 
Table 3: Comparison of training & testing results 
 
Splitting Ratio 
N 
R-squared 
RMSE 
MAE 
Training Results 
70% 
2097 
0.8178 
0.2668 
0.1969 
Testing Results 
30% 
900 
0.8034 
0.2673 
0.1970 
 
The hedonic regression model is performant and stable, exhibiting this quality in both the training and testing 
datasets. The model was developed from a 70-30 split of data, with 2,097 observations in the training set and 
900 observations in the testing set, thus giving a respectable number of observations for analysis. The 
measurement metrics shown in Table 3 also support the model's prediction ability. RMSE shows good agreement 
on both training (0.2668) and testing (0.2673) datasets, while the MAE remains stable across training (0.1969) 
and testing (0.1970) phases. The near-match of the error statistics between training and testing indicates that the 
model generalized well, without overfitting, making it a reliable tool for price prediction. Regarding R-squared, 
there are little decrease values for the model when measured on the testing set (with R-squared moving from 
0.8178 to 0.8034), indicating that the model's power didn’t have a significant decrease when confronted with 
new observations property which is highly desirable in any real estate valuation model. 
3.3.ANN results 
In this study, it must be researched how to tune the number of hidden layers and the number of nodes in each 
layer. The cross-validation process has been performed to determine them effectively. One or two hidden layer 
options have been considered. The number of nodes for each of the hidden layers is {5,10,15,20}. The literature 
has described several activation functions. The sigmoid activation function is the most commonly used one and 
is used in the present study as well. RMSE, R-squared, and MAE performance measures for each combination: 
one hidden layer or two hidden layers, four possible hidden layer node numbers based on training and testing 


===== PAGE 11 (score 29, numeric-paren 0, table-words 10, var-words 3) =====

36 
 
 
 
 
 
data sets. By these criteria, the best configuration was found using a sigmoid activation function, hidden nodes. 
The five and fifteen hidden nodes are hidden layer 1 and hidden layer 2, respectively. The lowest testing RMSE 
value was calculated using this configuration. The output is given in Table 5 for each combination. The model 
has been fitted using these options against the entire training data set. Testing results were thus achieved via this 
model using the testing dataset. To mitigate overfitting in the ANN model, we implemented L2 regularization 
(λ = 0.01) and applied a dropout rate of 20% in the hidden layers. Additionally, early stopping was used to halt 
training when validation loss did not improve for 10 consecutive epochs. Predictive neural network architectures 
for house price prediction analyses have been concluded to unveil fascinating structures in various 
configurations. The results comprise arguments around combinations of hidden layers and node configurations 
short among one-layer and two-layer networks that each have sigmoid function activation. Performance metrics 
RMSE, R-squared, and MAE illustrate noticeable variations across different architectures. As shown in Table 4, 
layer 1 is for single layer configurations, in which case 20 hidden nodes in its architecture gave the best 
performance compared to others in a one-layer fashion with an RMSE of 0.260158562 and R-squared of 
0.826944395, and an MAE of 0.193798218. Presumably, investigation indicates that increasing the number of 
nodes in a single layer generally improves model performance to a certain limit. Layer 2 is for two-layer 
configurations; optimal performance was obtained with the first hidden layer, having 20 nodes, and the second 
hidden layer with 5 nodes. Results were the best overall: RMSE of 0.258084424, R-squared of 0.829692798, 
and MAE of 0.190277261, making it the strongest performance among all applications. There are ranges of 
improvements over single-layer architectures; differences are small, and hence they indicate that using a second 
layer with a proper configuration of nodes may enhance the model's capability of prediction. Interestingly, more 
nodes in the second layer did not always lead to better performance, as found by the fact that larger second-layer 
configurations had slightly worse performance for single-layer architectures. This suggests that a more compact 
second layer might be better at capturing the underlying patterns in housing prices without overfitting. All 
configurations have high R-squared values (consistently above 0.82); hence, both single- and two-layer 
architectures explain a considerable portion of the variance in housing prices, while the two-layer structure is 
slightly ahead in terms of predictive accuracy, see Table 4. 
 
Table 4: ANN results 
Layer 
Act. Function 
#Hidden 
Nodes 1 
#Hidden 
Nodes 2 
RMSE 
R-squared 
MAE 
 
 
5 
- 
0.261197515 
0.825559427 
0.192662644 
1 
sigmoid 
10 
- 
0.260923272 
0.825925541 
0.191840764 
 
 
15 
- 
0.263006816 
0.823134376 
0.194910972 
 
 
20 
- 
0.260158562 
0.826944395 
0.193798218 
 
 
 
5 
0.266138597 
0.818897206 
0.199709806 
 
 
5 
10 
0.265614815 
0.819609354 
0.196885088 
 
 
 
15 
0.265299801 
0.82003698 
0.19616688 
 
 
 
20 
0.261160046 
0.825609471 
0.192564723 
 
 
 
5 
0.264051023 
0.821727181 
0.196315661 
 
 
10 
10 
0.262675799 
0.823579297 
0.194912253 
 
 
 
15 
0.260063996 
0.827070181 
0.192583763 
2 
sigmoid 
 
20 
0.26194096 
0.824564996 
0.192689893 
 
 
 
5 
0.259715786 
0.827532956 
0.193317165 
 
 
15 
10 
0.258833053 
0.828703341 
0.19196207 
 
 
 
15 
0.258158948 
0.82959443 
0.190836668 
 
 
 
20 
0.264755273 
0.820774971 
0.197949421 
 
 
 
5 
0.258084424 
0.829692798 
0.190277261 
 
 
20 
10 
0.261053266 
0.825752046 
0.193597329 
 
 
 
15 
0.261140706 
0.825635298 
0.195061518 
 
 
 
20 
0.263808809 
0.82205409 
0.195502868 
 
The neural network model demonstrated exceptional performance with extraordinary levels of predictive 
power for both training and testing datasets. The model was trained with 2,097 observations, while 900 


===== PAGE 12 (score 21, numeric-paren 0, table-words 8, var-words 3) =====

37 
 
 
 
 
 
observations were subjected to testing, using a 70%-30% split ratio, allowing for evaluation on more solid 
ground. From the training, the model's performance shows that the R-squared value was 0.9042. In effect, it 
explains 90.4% of the variance in housing prices in the training data. Together with this finding are very low 
error measures of RMSE: 0.1546 and MAE: 0.111, thereby confirming a high degree of accuracy for the 
predicted value during training, see Table 5. 
 
Table 5: Multilayer perceptron results 
 
Splitting Ratio 
N 
R-squared 
RMSE 
MAE 
Training Results 
70% 
2097 
0.9042 
0.1546 
0.1111 
Testing Results 
30% 
900 
0.8968 
0.2008 
0.1450 
 
Some potential decrease in performance was anticipated when the model was applied to the testing dataset; 
however, its performance remains strong. The testing R-squared of 0.8968 states that the model explains almost 
90% of the price variance in the unseen data. There is a moderate increase in error metrics during the testing 
phase, with RMSE currently at 0.2008 and MAE at 0.1450. Therefore, the variances in training and testing 
metrics indicate some given degree of overfitting, but otherwise, the model is successful in its predictive power 
applied to new data. The relatively small gap difference between training and testing performance indicates that 
the model has reached a good trade-off between fitting the training data while generalizing to new observations, 
thereby qualifying as a trustworthy housing price prediction tool. Regularization and early stopping help to offset 
the small overfitting shown by the train-test R2 gap (0.9042 vs. 0.8968), guaranteeing robust generalization. To 
minimize the overfitting issue, L2 regularization (λ = 0.01) and 20% dropout rates have been used on hidden 
layers. From sensitivity tests with λ = {0.001, 0.1}, the decrease in RMSE was minimal (<0.002), while 
increasing dropout to 30% raised test RMSE to 0.2051. Hence, 20% became the optimal dropout rate. Early 
stoppage occurred after 10 epochs, where validation loss ceased to improve. This brought the train and test R² 
that were previously 0.015 apart to just 0.0074. 
 
Figure 3. Observed and predicted house prices by multilayer perceptron. 
 
Figure 3 Compared with hedonic regression, ANN predictions come much closer to the actual, particularly 
in terms of midrange and higher properties. 
3.4.RF Result  
This machine learning method, which is very popular in the applications of ensemble learning, was used to 
fit the data for non-linear capture or non-linear relationships present in the real estate price data. It uses 500 
decision trees to construct an ensemble of trees, with each tree built differently through controlled complexity 
parameters (max depth=10, min samples split=5, min samples leaf=3) to avoid overfitting while still retaining 
the model's prediction power. The model was trained on 70% of the standardized dataset, of which the remaining 
30% was left for validation purposes. Random Forests would also be a better investment for real estate valuation 
because the complex interactions that take place with property and characteristics can be modeled without the 


===== PAGE 13 (score 30, numeric-paren 5, table-words 8, var-words 6) =====

38 
 
 
 
 
 
need for explicit specification. This was unlike traditional hedonic pricing models: the Random Forests method 
for threshold effects and segmentation patterns in the housing market, with certain features bearing different 
impacts at different price points. The built-in feature importance mechanism in the model gives insights into 
which factors cause the highest variations in property values, reinforcing an alternative to the coefficient-based 
way that hedonic regression explains the differences in value and Shapley Additive Explanations (SHAP) values 
from neural networks. The strong performance of the random forest model which trained as the same as Hedonic 
Regression and ANN on a 70/30 split utilizing 2,097 training samples and 900 testing samples is evident from 
R-squared values of 0.9315 (training) and 0.9065 (testing), which means that over 90% of variance in the target 
variable is explained by the model while generalizing well to new unseen observations. The RMSE (0.1577 
training, 0.1911 testing) and MAE (0.1116 training, 0.1353 testing) are low, indicating precise predictions, with 
only small increases in error on the test set, the result shown in Table 6. 
 
Table 6: Performance Metrics for Random Forest Model 
 
Splitting Ratio 
N 
R-squared 
RMSE 
MAE 
Training Results 
70% 
2097 
0.9315 
0.1577 
0.1116 
Testing Results 
30% 
900 
0.9065 
0.1911 
0.1353 
 
The feature importance scores in Table 7, generated by the random forest model, indicate that geographical 
location strongly dominates the prediction of property values in Riyadh. For example, "North of Riyadh" has an 
importance score of 0.768659, substantially higher than other features. This dominance implies that location 
variables heavily influence the model predictions, potentially overshadowing other structural and attribute 
variables, which should be considered when interpreting the model's robustness. 
 
Table 7: Feature importance by random forest 
 
Feature 
Importance 
11 
North of Riyadh 
0.768659 
14 
West of Riyadh 
0.612986 
13 
East of Riyadh 
0.589257 
12 
South of Riyadh 
0.283161 
0 
Building Type 
0.180799 
9 
Entrance Number 
0.171134 
5 
Size 
0.127103 
15 
Central Riyadh 
0.08131 
3 
Bathroom 
0.079826 
10 
Elevator 
0.057982 
4 
Kitchen 
0.049231 
2 
Living Room 
0.044905 
8 
Age 
0.038062 
1 
Bed Room 
0.037562 
7 
Floor 
0.028789 
6 
Street Wide 
0.023201 
 
Conversely, "Central Riyadh" sits at the other end with a lower importance of 0.08131, further hinting that 
it may play a less influential role in this regard. Some property-related parameters like "Building Type" 
(0.180799), "Entrance Number" (0.171134), and "Size" (0.127103) also have a noteworthy contribution, though 
certainly much less than location. Features such as "Bathroom," "Kitchen," "Living Room," and "Bed Room" 
have less than this importance (from 0.079826 to 0.037562), while "Street Wide" (0.023201) and "Floor" 
(0.028789) seem to be the most trivial. This distribution indicates that, in this model, spatial factors take 
precedence over structural or amenity-based characteristics when it comes to influencing predictions. 
 


===== PAGE 14 (score 7, numeric-paren 5, table-words 0, var-words 2) =====

39 
 
 
 
 
 
 
Figure 4: Observed and predicted house prices by RF. 
 
Figure 4 Most of the values predicted by the RF model closely match the actual housing prices, therefore, 
evidence of predictive power in this model can be concluded. The results also portray that RF is well suited for 
modeling nonlinear relationships and spatial effects, which are components that diminish the performance of 
both hedonic regression and ANN. Figure 5 The SHAP value chart indicates that location related variables (for 
example, distance from North Riyadh) top the list in affecting property prices, with property size and structural 
attributes (such as number of entrances) also contributing, and street width and number of kitchens having 
minimal impact.  
 
 
Figure 5: SHAP plot for feature impact 
 
The gradient color indicates the direction of the impact, where higher values for key features lead to 
increased property prices. The SHAP summary plot shows that location features "North of Riyadh" (0.768659) 
and "West of Riyadh" (0.612986) had the strongest influence on house price predictions; the higher the value of 
these features, the more value is added due to the proximity of amenities and infrastructures. "Size" (0.127103) 
also plays an important role, whereby larger properties increase values, while features like "Kitchen" (0.049231) 
and "Street Wide" (0.023201) remain irrelevant. The color gradient indicates non-linear effects whereby higher 
values of key features (location, size, etc.) will induce positive price consequences, in line with what was found 
in this study concerning location and structural attributes. 
 


===== PAGE 15 (score 38, numeric-paren 7, table-words 14, var-words 1) =====

40 
 
 
 
 
 
3.5.Comparison of the methods 
A comparison of the test performances of hedonic regression, ANNs, and RF is shown in Table 8. In terms 
of measures of performance R-squared, (RMSE), and Mean Absolute Error (MAE) both machine learning 
techniques generally have performance that ranks far above that of conventional hedonic regression: RF, indeed, 
has the highest predictive capacity overall. RF achieved the greatest R-squared (0.9065), or 91% explanation of 
variance in house prices, another side 90% and 82% for ANN and hedonic regression, respectively. And it 
produced the minimum prediction error, according to RMSE 0.1911 (compared with 0.2008 for ANN and 0.2668 
for hedonic regression) and MAE 0.1353 (compared with 0.1450 for ANN and 0.1969 for hedonic regression). 
 
Table 8: Comparison of test performance of hedonic regression, ANN, and RF 
 
Hedonic 
Regression 
ANN 
RF 
R-squared 
0.8034 
0.8968 
0.9065 
RMSE 
0.2673 
0.2008 
0.1911 
MAE 
0.1970 
0.1450 
0.1353 
 
Because both machine learning methods used here captured significantly non-linear relationships, they 
excelled over ordinary hedonic regression. RF, however, on all metrics, gave marginally better performance 
when compared to ANN. This shows that it's more efficient in modeling the complexity of Riyadh's housing 
market compared with single tree architectures like those of ANNs, hence, it's at a little higher precision level 
than ANNs, yet interpretable. In Table 9, a sample of 20 observations is presented of predicted house prices 
using hedonic regression and artificial neural networks (ANN) in Riyadh, Saudi Arabia, in log form and 
compared against the actual prices. The results show that among the three models, the RF model gave the best 
predictive feature, whereas the ANN and Hedonic Regression models were inferior. The higher R-squared value 
(0.9065) and lower RMSE (0.1911) and MAE (0.1353) values accounted for this and showed that RF could 
predict house prices more accurately and reliably than others. RF gave the best performance due to its capability 
to capture complex interactions and nonlinearities in the data. ANN is better in performance than Hedonic 
Regression since it has an R-squared value of 0.8968, showing that it can identify nonlinear patterns, though it 
has a higher RMSE and MAE than RF. Hedonic Regression, though the traditional method, is the most apparent 
of the three, with an R-squared of 0.82, indicating that the model might not capture the full real estate price 
determinant complexity. 
 
Table 9 above shows that among the three models, the RF model gave the best predictive feature, whereas 
the ANN and Hedonic Regression models were inferior. The higher R-squared value (0.9065) and lower RMSE 
(0.1911) and MAE (0.1353) values accounted for this and showed that RF could predict house prices more 
accurately and reliably than others. RF gave the best performance due to its capability to capture complex 
interactions and nonlinearities in the data. ANN is better in performance than Hedonic Regression since it has 
an R-squared value of 0.8968, showing that it can identify nonlinear patterns, though it has a higher RMSE and 
MAE than RF. Hedonic Regression, though the traditional method, is the most apparent of the three, with an R-
squared of 0.82, indicating that the model might not capture the full real estate price determinant complexity. 
Figure 6 brings up the fact that RF produces the best prediction as compared to ANN, while hedonic regression 
significantly underestimates prices. This also indicates that the ML models yield better results owing to the 
strong influence of nonlinear and spatial effects in the real estate market of Riyadh. The results demonstrate 
ANN's effectiveness in modeling nonlinear relationships and interactions among features (e.g., Size, Location), 
likely enhanced by cross-validation during model development. Hedonic regression, although providing a more 
robust structure, appears less adept at sorting heterogeneous and varying high-valued properties, primarily 
influenced by unique urban factors of Riyadh, such as congestion and access to amenities. These takeaways from 
the choice observations reveal that ANN extended improved predictive accuracy throughout the 2,997-property 
dataset, suggesting it could be a more effective tool for real estate valuation in this regard. 
 
 
 
 
 
 


===== PAGE 16 (score 12, numeric-paren 0, table-words 1, var-words 2) =====

41 
 
 
 
 
 
Table 9: Predicted house prices obtained by hedonic regression, ANN, and RF 
Cases 
Actual Price 
Hedonic Regression Price 
ANN prices 
RF 
1. 
12.67607627 
12.97913951 
14.02735233 
12.79326833 
2. 
13.44444688 
13.09504317 
13.32105923 
13.39909212 
3. 
13.50079981 
13.50260718 
13.51704311 
13.40802153 
4. 
15.36307307 
15.19602075 
14.73604679 
15.20827409 
5. 
14.15198279 
13.46822121 
13.66140079 
13.92980437 
6. 
14.25376549 
13.58558079 
14.02284718 
14.00800457 
7. 
14.45208739 
14.11736312 
14.19547176 
14.14799829 
8. 
15.36307307 
15.19602075 
14.73604679 
15.20827409 
9. 
15.52025865 
15.25070181 
14.63071537 
15.4457558 
10. 
13.01700286 
12.99152837 
13.32105923 
12.97964453 
11. 
13.54107371 
13.70104505 
13.49339771 
13.76861773 
12. 
13.57978822 
13.41867479 
13.42337124 
13.49201393 
13. 
13.59236701 
13.98696611 
13.73884885 
13.65702343 
14. 
13.61705962 
13.98896421 
13.89270578 
13.60157394 
15. 
13.67394699 
13.76853638 
13.68364811 
13.79487572 
16. 
13.68767719 
14.04977926 
14.08522486 
13.69303799 
17. 
13.76421726 
14.06136506 
13.92757988 
14.05108602 
18. 
13.81551056 
13.96331322 
13.845191 
13.84581778 
19. 
13.99783211 
14.06805398 
13.95209458 
14.00282764 
20. 
14.07787482 
13.65837871 
13.90108395 
13.84240501 
 
 
 
Figure 6: Actual and predicted prices by hedonic regression, ANN, and RF 
Figure 6 reveals distinct prediction patterns across models. Hedonic regression systematically 
underestimates high-value properties, with predictions deviating up to 15-20% below actual prices for properties 
exceeding SAR 3,000,000. This pattern suggests the linear functional form fails to capture exponential price 
premiums associated with luxury features and prime North Riyadh locations. 
ANN predictions show improved performance in the mid-range, reducing underestimation to 8-12%, but 
still exhibit scatter for properties above SAR 4,000,000 (log price ~15.2). This indicates the neural network 
captures some nonlinear relationships but may require deeper architectures or additional training for ultra-high-
value properties. 
RF demonstrates superior performance across the entire price spectrum, with minimal deviation even for 
properties exceeding SAR 5,000,000 (log price ~15.4). The ensemble approach effectively models the complex 
interactions between location (particularly North Riyadh premium), property size, and luxury features that