Conference PaperPDF Available

A Survey on Application of Artificial Intelligence in Real Estate Industry


Abstract and Figures

This paper will discuss the use of Artificial Intelligence (AI) in real estate industry. Today, besides Multiple Regression Analysis (MRA) models the use of AI systems for real estate valuation becomes better alternative. These AI systems for real estate valuation are more recent and becoming practical. Even if there are a number of artificial intelligent systems, Artificial Neural Networks (ANN) and Expert Systems (ES) are the ones presently applied for real estate valuation. Thus this paper will examine the current trends of ANN and ES and considers suitable applications in real estate valuation. In addition, prediction capability comparison between ANN and MRA will be presented by considering different case studies since both use statistical analysis and data modelling. Furthermore, common characteristic of ANN and ES will be compared. Beside ANN and ES, this paper will also discuss the application of hybrid systems for real estate valuation which mitigate the limitations and take advantage of the opportunities to produce systems that are more powerful than those that could be built with single intelligent systems.
Content may be subject to copyright.
A Survey on Application of Artificial Intelligence in Real Estate Industry
Woubishet Zewdu Taffese
Rehtorinpellonkatu 4B 405, 20500 Turku, Finland
Tel: +358 40 8749981,
This paper will discuss the use of Artificial Intelligence (AI)
in real estate industry. Today, besides Multiple Regression
Analysis (MRA) models the use of AI systems for real estate
valuation becomes better alternative. These AI systems for
real estate valuation are more recent and becoming
practical. Even if there are a number of artificial intelligent
systems, Artificial Neural Networks (ANN) and Expert
Systems (ES) are the ones presently applied for real estate
valuation. Thus this paper will examine the current trends
of ANN and ES and considers suitable applications in real
estate valuation. In addition, prediction capability
comparison between ANN and MRA will be presented by
considering different case studies since both use statistical
analysis and data modelling. Furthermore, common
characteristic of ANN and ES will be compared. Beside
ANN and ES, this paper will also discuss the application of
hybrid systems for real estate valuation which mitigate the
limitations and take advantage of the opportunities to
produce systems that are more powerful than those that
could be built with single intelligent systems.
Keywords: Real Estate Valuation, Artificial Intelligence,
Artificial Neural Networks, Expert Systems, Hybrid
An accurate and fast prediction of the real estate value is
important to prospective homeowners, developers,
investors, appraisers, tax assessors and other real estate
market participants, such as, mortgage lenders and insurers.
Real estate valuation based on traditional approaches such
as cost and sale comparison approach lacks an accepted
standard and a certification process. Therefore, the
availability of a real estate value prediction model helps to
fill up an important information gap and improve the
efficiency of the real estate market [5].
Over the last two decades there has been a proliferation of
empirical studies analyzing residential real estate values.
The use of computer for real estate valuation began in the
early 1980s, coinciding with the development of
information systems technology. Subsequently, different
statistical techniques were incorporated to process market
data, among which the method of MRA proved especially
relevant [14]. MRA models are the most popular
quantitative technique in real estate valuation. It has been
applied in various residential real estate valuations to assist
appraiser in statistical analysis and complement the
traditional sales comparison approach. MRA methods have
experienced criticism from the academic and practitioner
community. MRA has often produced serious problems for
real estate valuation that primarily result from
multicolinearity issues in the independent variables and
from the inclusion of “outlier” properties in the sample.
Moreover, nonlinearity within the data may make multiple
regressions an inadequate model for a market that requires
precise and fast responses.
Nowadays, besides MRA models the use of AI systems for
real estate valuation becomes better alternative. Using AI
systems for real estate valuation is more recent and
becoming practical. Since then there have been numerous
experiences, and the creation of new models is on the
increase. Even if there are a number of AI systems, ANN
and ES are presently applied for real estate valuation.
Artificial Neural Network
An Artificial Neural Network (ANN), also called a Neural
Network, is an interconnected group of artificial neurons
that uses a mathematical or computational model for
information processing based on a connectionist approach
to computation. There is no precise agreed definition
amongst researchers as to what an ANN is, but most would
agree that it involves a network of relatively simple
processing elements, where the global behaviour is
determined by the connections between the processing
elements and element parameters. The original inspiration
for the technique was from examination of bioelectrical
networks in the brain formed by neurons and their
synapses. In an ANN model, simple nodes (called variously
“neurons”, “neurodes”, “processing elements (PEs)” or
“units”) are connected together to form a network of nodes
hence the term "neural network”.
ANNs usually have several layers. The first layer is called
the input layer, the last one the output layer. The
intermediate layers (if any) are called the hidden layers
which can not be inspected from the outside. Example of
simple ANN structure is shown in Figure 1.
Figure 1 - Simple example of ANN
The information to be analyzed is fed to the neurons of the
first layer and then propagated to the neurons of the second
layer for future processing. The result of this processing is
then propagated to the next layer and so on until the last
layer. Each unit receives some information from other units
Layer Output
Proceedings of the Third International Conference on Artificial Intelligence in Engineering & Technology
November 22-24, 2006, Kota Kinabalu, Sabah, Malaysia
or from the external world and processes this information,
which will be converted into the output of the unit.
Depending on the desired functionality and problem area,
ANN can be likened to the way
te Valuation
Figure uation
Different researches show that ANNs for real estate
Based on ANN real
Table 1 – ANN vs. MRA Performance comparisons
Researchers Data Place ANN vs. MRA
neurons can be structured in a number of different
architecture. In general it is possible to distinguish three
main types of network architecture: single-layer
feedforward networks, multi-layer feedforward networks
and recurrent networks [1].
The learning process of the
a child learns to recognize patterns, shapes and sounds, and
discerns among them. In real neurons the synaptic strengths
may, under certain circumstances, be modified so that the
behaviour of each neuron can change or adapt to its
particular stimulus input. In artificial neurons the equivalent
of this is the modification of the weight values. ANNs never
work the first time round. Thus, they need to “learn”. With
standard learning algorithms an ANN learns through an
iterative process of weight adjustment. The type of learning
is defined by the way in which the weights are modified.
The three main learning paradigms are: 1) supervised
learning, 2) unsupervised learning, and 3) reinforcement
learning [1]. Finally, testing can be done either from
randomly selected learning set or from a set of observations
immediately following the learning set.
Implementation of ANN for Real Esta
ANNs are often used for a statistical analysis and
modelling, in which their role is perceived as an alternative
to standard nonlinear regression or cluster analysis
techniques. Thus, they are typically used in problems that
may be couched in terms of classification or forecasting.
Forecasting is primarily a quantitative process using
numerical data from the past to forecast the future. Since
real estate valuation is forecasting process ANNs can be
used for real estate valuation [17].
2 - Simple model of ANN for Real Estate Val
valuation usually work in the range of 10 to 50 variables
[14]. It features an input layer with the same number of
neurons, likewise the second hidden layer (although this
can vary between half and double the number of variables),
and an output layer containing a single neuron.
Feedforward ANNs are most commonly trained using a
back-propagation algorithm for training and have been
widely used for several civil engineering applications [4].
Indeed feedforward /back-propagation network models are
the most common form of ANN models used for real estate
valuation. The majority of ANNs designed for real estate
valuation is similar in structure with the structure in Figure
2. ANN algorithms typically begin with randomly
determined or equal default weights for each of the nodes in
each of the hidden layer(s). In each model-training, each
real estate attribute is entered into the model, the network
sums and transforms the values of the input variables into
the predicted output value(s). The model then compares the
ANN’s estimated price to the actual price. If a discrepancy
exists, then the software works backwards to adjust the
hidden layer weights to minimize the prediction error.
These adjustments are similar in nature to the adaptive
estimation techniques used in MRA for real estate
valuation. While training, ANN models repeat these steps as
the data for each new real estate are added, always
adjusting the hidden layer weights to minimize the total
prediction error. ANN stops training when it reaches a
preset internal error threshold, either the software’s default
error level or the researcher’s pre-designated error
threshold. Such a threshold is needed because without one,
an ANN would effectively memorize, or “over-train” on the
training data, and its predictive ability towards a new real
estate would significantly deteriorate.
Case-Studies of Real Estate Valuation
Effort to apply ANN technology to the valuation of
estate is dated from the early 1990s. Frequently these
studies are in the form of comparative analysis, with
researchers contrasting the findings and perceived
efficiency of ANN models with more tried and tested
statistical methods. Given the potential difficulties
associated with regression modelling, namely functional
form and non-linearity of variables, ANNs have found a
measure of insightful appeal [13].
Lot Area
r Ar
Age of Do et al. [2] 105 ---- ANN 2× accurate
T Singapore A ay et al. [19] 833 NN 1.92× accurate
Evans et al. [3] 34 England
& Wales ANN better
The studies shown in Ta nerall rt the ble 1 ge y suppo
superiority of ANN over MRA in predictive ability. There
are also studies showing the superiority of MRA over ANN
while other studies show inconclusive results. For example
Worzala et al. adopt a contrary position and cast some
doubt upon the role of neural networks compared with
MRA models, suggesting that caution is needed when
working with neural networks. In undertaking analysis at
varying levels of investigation and utilizing different neural
network shells, the error magnitude for individual
properties was found in some cases to be very significant
and clearly not acceptable for a professional appraisal [20].
Other researchers McGreal et al. also adopts a more
sceptical approach to the potential merits of neural
networks within the valuation process and in this respect
Number of
Quality of
Va l ue
Proceedings of the Third International Conference on Artificial Intelligence in Engineering & Technology
November 22-24, 2006, Kota Kinabalu, Sabah, Malaysia
agrees with the cautionary tone expressed by Worzala et al.
with reference to the position taken by other researchers
Different recent researches show that neural networks are
training size
NN for Real Estate Valuation tability,
their lack of transparency.
Expert System
ES has a wide base of knowledge in restricted domain, and
ural components with
BR for Real Estate Valuation
Initially the expert will provide the heuristic knowledge
nted in the real estate valuation
found to perform better than MRA when either the number
of groups or the number of variables increases and also
when the classification tasks tends to become complex [9].
Similarly study on real estate valuation gives a plausible
explanation why previous studies have obtained varied
results when comparing MRA and ANN predictive
performance for real estate values. The predictive
performance depends on the evaluation criteria used in
combination with the training size and model specification.
Fluctuation in the ANN model’s performance may be due to
the larger number of parameter settings chosen via
experimentation and dependent on training sample size
[15]. Also due to lack of some environmental attributes
relationship between real estate attributes and real estate
price is non-linear, thus it could be the cause of the poor
performance of the MRA models. Conversely, the ANN
model can overcome some of the problems related to the
data patterns and the underlying assumption of the MRA
model. As a result, the model yields a better prediction
result when compares with MRA model [11].
In conclusion, if one provides sufficient data
and appropriate ANN parameters, then ANN performs
better than MRA.
Pros and Cons of A
The most advantageous property of ANN is its adap
which allows the neural network to perform well even when
the environment or the system being controlled varies over
time. Thus using ANNs for real estate valuation is
advantageous because it has time dependent attributes.
ANNs learn system behaviour by using system input-output
data and do not require update when input changes. The
learning and generalization capabilities of ANNs enable it
to more effectively address nonlinear, time variant
problems, even under uncertain or erroneous attributes of
real estate valuation. Thus, ANNs can solve problems that
either unsolved or inefficiently solved by traditional real
estate valuation techniques. It can also develop solutions to
meet a pre-specified accuracy.
A major disadvantage of ANN is
The internal structure of the network is hidden and may not
be easy to duplicate, even using the same data inputs. This
leads to lack of accountability because the system’s
intermediate steps can not be checked [9]. Furthermore it is
difficult to determine the proper size and structure of ANN
which determines the value of real estates.
uses complex inferential reasoning to perform tasks which a
human expert could do. It encompasses several different
components such as a knowledge base, inference
mechanisms, explanation facility, etc. All these different
components interact together in simulating the problem
solving process by acknowledged expert of a domain. They
are based on knowledge to solve problems that would
normally require a human expert. The knowledge is
collected from human experts and secondary knowledge
resources, such as books, and is represented in some form,
often using logic or production rules. The system includes a
reasoning mechanism as well as heuristics for making
choices and navigating around the search space of possible
solutions. It also includes a mechanism for passing
information to and from the user.
ES has four major architect
individuals in various roles. These are The Knowledge
Base, Working Storage, Inference Engine, and User
Interface. In ES model, knowledge can be represented and
stored in the knowledge base in various forms. For
example, one of the most commonly used ways to represent
knowledge is rule-based reasoning, in the form of IF-THEN
rules. But using rule-based reasoning in real estate
valuation could be difficult to implement because rules
would be based on observations of the market, which is a
dynamic entity. Thus, the rules would likely have to be
updated frequently after studying market data. An
alternative of rule-based reasoning is Case-Based
Reasoning (CBR). A CBR solves new problems by adapting
solutions that were used to solve old problems. It is most
useful in knowledge domains where precedence-based
reasoning is appropriate. Domains such as medical
diagnosis and audit commonly use CBR. Nowadays, real
estate valuation lends itself well to CBR. Since the method
to be followed is the market data approach, the case library
will consist of descriptions of all kinds of real estates
previously sold [6].
Implementation of C
CBR is built on the premise that humans use an analo
or experiential reasoning approach to learn and to solve
complex problems. It involves two primary steps: (1) find
those cases in storage that have solved problems similar to
the current problem, and (2) adapt the previous solution(s)
to fit the current problem context. Figure 3 shows nominal
process of CBR.
Figure 3 – Nominal process in CBR
necessary to adapt or adjust the values from the real estate
properties in the case library that are the most similar in
terms of characteristics or features to the real estate being
appraised. To determine the solution it is necessary to
follow the following four major steps:
Step 1: Problem Input
The cases to be represe
Input Case
Retrieval Case
A n daptatio Tes t
Adaptation Similarity
Metrics Methods
Proceedings of the Third International Conference on Artificial Intelligence in Engineering & Technology
November 22-24, 2006, Kota Kinabalu, Sabah, Malaysia
model consist of the descriptions of real estates sold during
a specific period of time in a certain geographic area. Even
when the elements of comparison may vary somewhat from
appraiser to appraiser and from market to market, a set of
elements of comparison should be chosen such as area in
square unit, number of bedrooms, number of bathrooms,
age of the house, location, architectural style of the house,
date of sale, type of HVAC equipment, type of garage, lot
size, etc.
Step 2: Case Retrieval tates that are similar to the subject
se Adaptation emory are identified, they are
is one
and Conclusion hey are added or
l Estate Valuation Based on CBR tate
on residential apartments in
for Real Estate Valuation s mostly
Characteristics Comparisons of ANN and ES
Even if ANN and ES use different approaches they have
In this step, three real es
property will be retrieved since real estate appraisers
choose only three comparable properties to be adjusted and
included in their valuation [6]. Valuation of real estate
based on CBR is particularly interested in the global best
matches with prior sales records. In the real estate valuation
model, the best-match algorithm takes the target case,
computes a similarity metric between the target case and
each case in memory and retrieves the best matches from
Step 3: Ca
Once the best cases in m
retrieved from memory to become the official comparable
properties. Then, the adaptation phase begins. This phase
applies adjustments to the sale price of comparables to get a
better indication of the value of the subject property.
Probably the best understood adaptation technique
called parameterized solutions. The parameter to be adapted
is determined by the differences between the subject
property and it’s comparable from the case library. When a
case is retrieved for an input situation, the old and new
problem descriptions are compared along the specified
parameters. The differences are then used to modify the
solution parameters in the appropriate directions. In real
estate valuation, the comparison parameters are used to
determine what adjustments are needed to “adapt” the sale
price of each comparable to the features of the subject
property. Thus, the solution parameters are the sale prices
of the comparables.
Step 4: Test Solution
Once all adjustments are obtained, t
subtracted, as appropriate, from the sale price of the
corresponding comparable property. In this way, an
adjusted value, which better reflects the value of the subject
property, is produced for each one of the three comparables.
Then, the new solution is tested and, if successful, added to
the case library. If, however, the test fails, then the
adaptation process must be revised or a new set of case
must be retrieved.
Case-Studies of Rea
Gonzalez et al. verified the ability of CBR for real es
valuation. The case library contains multiple listing service
manual with description of 107 sample of single-family
residential real estates sold during a period of about five
months in the area of Deltona, FL. The prototype number of
elements of comparison was only 12. However, their CBR
model appraise values were fairly consistent and close to
the list prices. If more features are included for each real
estate, a better differentiation between real estates can be
made. Therefore, it is concluded that the values appraised
would be more precise [6].
Based on the study done
Bangkok, Thailand, Pacharavanich et al. concluded that,
CBR real estate valuation system has potential to become a
viable commercial tool for the valuation of residential real
estate in Bangkok. Appraisers also confirm that the CBR
model is easy to use, its usefulness and confidence for the
conclusion. Confidence in the conclusions provided by the
artificial intelligence systems reflects attitudes that users
can have towards the system after using it in training or for
problem solving [16].
Pros and Cons of CBR
In real estate valuation sales comparison approach i
used among other approaches. Traditionally in sales
comparison approach collection and retrieval of data on
sales of similar properties and adjustment of the sales data
are performed either manually or using non intelligent
systems which makes the process tedious. On the other
hand in CBR the system search for the most similar
properties from library, retrieve it and make the adjustment
easily. This takes a very short time to finalize the valuation
process, leads to cost reduction. The accuracy of the
appraised value is higher in CBR systems because it is
easier to find the best match for the subject property. Also it
reduces the required number of expertise significantly for
real estate valuation. Since these systems are based on
knowledge it can provide an accessible, available
alternative that can be used as a training tool for beginners.
As concluded by the above two case studies using CBR
systems results in decreased costs for appraisers, reduced
downtime and increased quality and throughput.
The knowledge may be internally consistent but in
due to either expert error or misunderstanding at the
acquisition stage. Thus it is not an easy task to develop
complete, consistent and correct knowledge bases. Some
tool support is available, and usage of a structured approach
can alleviate the problem. Therefore the appraised value
may be incorrect due to the internal error of the system.
Furthermore it requires a number of similar subject
properties already stored in the case library to reach on the
best match since the CBR system for real estate valuation is
more dependent on the case library.
some common characteristics which can be used to
compare them [10]. ANN has various advantages and
disadvantages compared with ES. From functional and
application standpoints, each approach can be equally
feasible, although in cases one may have an overall
advantage over the other. As we already discussed. ANN
and ES follow two quite different approaches to evaluate
the real estate. They have different properties, advantages
and disadvantage with regard to this application. ANN is
based on numeric computations and algorithms, while ES is
based on symbolic and heuristic reasoning. ANN has
capabilities of association, memorization, error-tolerance,
self-adaptation, and multiple complex pattern processing.
On the other hand, they cannot explain their own reasoning
behaviours and cannot adapt new environment (those not
made available previously in training the networks).
Whereas ES has the ability to explain their reasoning
behaviours and can adapt new environment using
knowledge bases. ES has obvious knowledge representation
forms that make knowledge easy to manage. However, self-
learning is still a problem and computation time can be
lengthy depending on the size of the domain and the range
of cases that must be realized. But ANN can analyze a
large number of cases quickly to provide accurate
responses. Though, validation of the content of ANN (i.e.,
the determination of the completeness and consistency of
the representation) is relatively more difficult than ES.
Table 2 – Characteristics of comparisons of AAN and ES
Characteristics ANN ES
Approach N Sy ic umeric mbol
Reasoning A ssociative Logical
Operations B e Mechanical iological-Lik
Fe Redback to user None easoning Path
P Parallel rocessing Approach Sequential
System Self ing -Organiz Closed
K nowledge Many examples Expert
S elf Learning Inherent None
Fault tolerance Tolerant Not
Maintenance Easy Difficult
AAda nd pr g
daptability to
changes ptive a
Flexible Requires
Lear city ning Capa High None
Characteristics comparison between, ANN and ES is shown
Hybrid of ANN and ES
The development of information systems based on the
d as useful technologies
years there has been an explosive growth in the
d of ANN and ES
Real estate valuation is no longer a traditional business that
in Table 2. The different characteristics of these systems
suggest that they can enhance each other to provide
solutions that neither of the system alone can deliver nor
lead to good solutions with less system complexity.
Because of these reason integrating or combining of two or
more intelligent systems become a primary concern for
researchers and practitioners. In the following section the
use of integrating these systems is presented in detail.
combination of two or more intelligent systems has been a
solution to overcome the limitations presented by individual
intelligent systems [8]. In the past decade, the amount of
research and development involving hybrid intelligent
systems has increased rapidly. Initial work addressed the
integration of ANN and ES [21].
ES and ANN are well establishe
that can complement each other in powerful hybrid systems
[7]. These integrated systems can also involve database and
other technologies to produce the best solutions to complex
problems. Due to their fundamentals natures, ANN and ES
have a natural synergism that can be exploited to produce
powerful computing systems. It becomes more efficient and
effective computing systems, making up for deficiencies in
the conventional approaches. A hybrid approach which
integrates ES and ANN has promising results to solve
problems in a fashion more consistent with human
intelligence. Interesting areas of research and development
include the use of ANN in situations where ES have
previously been used, development of application models
and guidelines when best to use hybrid systems, and further
work on creating development tools and environments.
Case-Study of Real Estate Valuation Based on Hybrid
In recent
successful use of hybrid intelligent systems in many diverse
areas. There are few ongoing researches on hybrid
intelligent systems which have been developed for real
estate valuation. For example a study shows the
performance of an integrated model named “Geo-
Information Neural System” (GINS) which is developed as
an alternative for use in the valuation of single-residential
real estate. This system integrates a Geographic Information
System (GIS) technique with ANN modelling [18]. GIS is
utilized for location distance measurements, spatial queries
and thematic mapping whilst ANN is employed to replicate
the way the human brain might process data by learning
relationships, in this case the one existing between property
characteristics such as physical and location attributes and
sales price. The results indicate that GINS provides an
efficient tool that provides superior residential real estate
valuations, while accuracy is improved by minimizing the
influence of subjective judgments. The other example of
hybrid system uses the hybrid of ANN and genetic
algorithms in association with a nearest neighbour
algorithm for mass valuation of real estate [12]. Though
ANN and genetic algorithms have relatively poor levels of
model transparency, the ability to objectively select and
retrieve comparables based on an integrated nearest
neighbour algorithm provides a solid foundation on which a
transparent mass appraisal system can be developed.
Optimum weights for the real estate attributes have been
derived from information inherent within the data, thus
negating the need to solely rely on subjective domain
knowledge to determine these. The application of enhanced
distance metrics such as mean, coefficient of variation and
significant mean clearly improve weight determination with
specific regard to categorical variables.
As of my knowledge there is no hybri
system which is developed for real estate valuation.
However using hybrid of ANN and ES is very promising
for real estate valuation as explained in this paper.
relies only on expert opinions of value. The profession is
now facing greater transformation in the valuation process
and methodology, along with innovations in information
technology. Technology is having profound effect on the
profession, as well as influence on the real estate valuation
Proceedings of the Third International Conference on Artificial Intelligence in Engineering & Technology
November 22-24, 2006, Kota Kinabalu, Sabah, Malaysia
process, largely pressured by the needs of today’s clients
who demand quick, easy and more objective process to
arrive at the opinion of value. The needs somehow motivate
dependency on intelligent valuation system that allows
clients to get faster and accurate value.
As shown in the case studies of this paper ANN and ES
[1] Castro, L.N. et al. eds. 2002. Artificial Immune Systems:
[2] . A Neural Network Approach to
[3] rtificial Neural Networks: An
[4] for
[5] ting the Value of Apartment
[6] zalez, A.J. et al. 1992. A Case-Based Reasoning
[7] ral Networks
[8] 2004. Advanced Topics in
[9] parison of Neural Networks
[10 ciety:
[11 al. 2004. House Price Prediction:
[12 of
[13 Neural Networks: The
[14 Internet sources
systems are already used successfully for real estate
valuation. Different researcher’s comparison result shows
that ANN gives better accuracy than MRA. However there
are few studies showing the superiority of MRA over ANN
while other studies reached on an inconclusive result.
Hybrid systems mitigate the limitations and take adva
of the opportunities to produce systems that are more
powerful than those that could be built with single
intelligent systems. The hybrid systems represent a range of
building blocks that may eventually allow simulating
human-like intelligence. Two successful hybrid systems
developed for real estate valuation is presented in this
paper. These hybrid systems demonstrate their feasibility
and advantages for single and mass real estate valuation.
Thus it is recommendable to use hybrid systems for real
estate valuation because it can reduce significantly the time
and cost required to value a real estate, while enhancing the
accuracy of value estimates.
A New Computational Intelligence Approach. Great
Britain: Springer.
Do, Q. et al. 1992
Residential Property Appraisal. The Real Estate
Appraiser 58(3): 38-45.
Evans, A. et al. 1992. A
Application to Residential Valuation in the UK. Journal
of Property Valuation & Investment 11(2) 195-204.
Flood, I. et al. eds. 1998. Artificial Neural Networks
Civil Engineers: Advanced Features and Applications.
USA: ASCE Publications.
Frew, J. et al. 2003. Estima
Buildings. Journal of Real Estate Research 25(1): 77-
Approach to Real Estate Property Appraisal. Expert
Systems with Applications 4(2): 229-246.
Kartam, N. et al. eds. 1997. Artificial Neu
for Civil Engineers: Fundamentals and Applications.
USA: ASCE Publications.
Khosrow-Pour, M. eds.
Information Resources Management, vol.3. Hershey and
London: Idea Group Inc.
Kumar, U.A. 2005. Com
and Regression Analysis: A New Insight. Journal of
Expert Systems with Applications 29(2): 424-430.
]Lan, Y.C. eds. 2005. Global Information So
Operating Information Systems in Dynamic Global
Business Environment. Hershey and London: Idea
Group Inc (IGI).
]Limsombunchai, V. et
Hedonic Price Model vs. Artificial Neural Network.
American Journal of Applied Sciences 1(3): 193-201.
]McCluskey, W. et al; 1999. The Application
Intelligent Hybrid Techniques for the Mass Appraisal of
Residential Properties. Journal of Property Investment
and Finance 17(3): 218-238.
]McGreal, S. et al. 1998.
Prediction of Residential Values. Journal of Property
Valuation and Investment 16(1): 57-68.
]Mora-Esperanza, J.G.4 June 2006. Citing
[15]Nguyen, N. et al. A. 2001. Predicting Housing Value:
[16 000. The Development of a
[17 P. 2000. Using Expert Systems and Artificial
[18 tegrating Artificial Neural
[19 lligence and the
[20 eural
[21 d
A Comparison of Multiple Regression Analysis and
Artificial Neural Networks. Journal of Real Estate
Research 22(3): 313-335.
]Pacharavanich, P. et al 2
Case-Based Reasoning System as a Tool for Residential
Valuation in Bangkok. In Proceedings of the Sixth
Annual Pacific-Rim Real Estate Society Conference,
Intelligence for Real Estate Forecasting. In Proceedings
of the Sixth Annual Pacific-Rim Real Estate Society
Conference, Sydney.
] Sarip, A.G. 2005. In
Networks and GIS for Single-Property Valuation. In
Proceedings of the Eleventh Annual Pacific-Rim Real
Estate Society Conference. Melbourne.
]Tay, D.P.H. et al. 1992. Artificial Inte
Mass Appraisal of Residential Apartments. Journal of
Property Valuation & Investment 10(2): 525-540.
]Worzala, E. et al. 1995. An Exploration of N
Networks and Its Application to Real Estate Valuation.
The Journal of Real Estate Research10 (2): 185–201.
]Zhang, Z., and Zhang, C. eds. 2004. Agent-Base
Intelligent Systems: An Agent-Based Framework for
Complex Problem Solving. Germany: Springer.
Proceedings of the Third International Conference on Artificial Intelligence in Engineering & Technology
November 22-24, 2006, Kota Kinabalu, Sabah, Malaysia
... Latshaw's (2019, 72-78) suggestions for CEOs may be summarized in terms of alertness to changes in priorities and readiness to be involved in new business activities (Latshaw, 2019, 72-78). Taffese (2016) explains that the real estate profession is responding to client demand for faster and more accurate property valuations. The author presents two AI valuation methods, Artificial Neural Networks (ANN) and Expert System (ES). ...
... ANN system's reasoning is difficult to ascertain, and results are often tentative. However, a trend toward the use of AI for property valuation has already begun (Taffese, 2016). ...
... The ANN technique can deal with non-linear relationships, allowing for a broader range of variations [21]. It can also adapt to changing surroundings, deal with noisy and ambiguous input and generalize previously unknown scenarios [22]. Comparing ANN and hedonic models showed that the ANN model predicted more accurately than the hedonic pricing model [23]. ...
Full-text available
The real estate land price valuation is a global issue, and its importance is not limited to the real estate market but also in the banking sector, insurance sector, and governing bodies for taxation and acquisitions. This paper compares the accuracy of the Hedonic Pricing Model (HPM) and Artificial Neural Network (ANN) model in predicting the residential land price for Chengalpattu district, a suburban region in the southern part of India. Residential land prices and data for the variables affecting land prices were collected and used to develop the HPM and ANN models. Subsequently, both models predicted land prices for newer land parcels, and their accuracy was compared. The performance evaluation indices of root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), R-square and accuracy were calculated for the predicted results of both the models. The HPM predicted the residential land prices with an accuracy of 75 %, whereas the ANN model predicted the prices with an accuracy of 91 %. The study showed that the ANN model is reliable and accurate in residential price prediction for suburban regions.
... AI or more widely known as AI is one of such technology which has improved the living standard of people [2] worldwide. AI is widely used now days in various fields like healthcare [3], real estate [4], stock market prediction [5], weather prediction [6], automobile [7] and also in many other fields [8] [9]. AI has many subfields like Natural Language Processing, Machine Vision, Robotics, Expert System etc [10], but in this study ML is used. ...
Conference Paper
Full-text available
Everyone wishes to buy and live in a house which suits their lifestyle and which provides amenities according to their needs. There are many factors that are to be taken into consideration like area, location, view etc. for prediction of house price. It is very difficult to predict house price as it is constantly changing and quite often the prices are exaggerated for which people who want to buy houses, and various real estate agencies who want to invest in properties, find it difficult to buy or sell houses. For this reason, in this paper the author creates an advanced automated Machine Learning model using Simple Linear Regression, Polynomial Regression, Ridge Regression and Lasso Regression using the Boston house dataset to predict house price in future accurately, and to measure the accuracy of these models various measuring metrics like R-Squared, Root Mean Square Error (RMSE) and Cross-Validation are used. This paper also studies the correlation of various attributes of the Boston dataset using the heat map to see which attributes actually impact the prediction of the models. It removes the outliers which are present in the dataset to achieve good accuracy. In this paper it is observed that Lasso Regression performs better in all the measuring matrices whereas Simple Linear Regression performs poor in all of the measuring matrices.
... However, to ensure the security and safety of the huge capital invested, grow expected on return, minimise risk, boost investor's confidence and pull investors across the globe, there is a need for a sustainable real estate market that is strongly driven by real-time data, technology innovations, ICT compliant and sophisticated market analytical model. However, one of the pragmatic approaches of achieving sustainable real estate market analyses is the application of artificial intelligence (AI) analytical tools which have been proved to be reliable, yielding a better and accurate results in the areas of market appraisal, trend analysis and forecasting (Taffese, 2006;Kabaivanov and Markovska, 2021). ...
Conference Paper
Full-text available
The information on the real estate market is fast growing and becoming more complex to analyse. Therefore there is a need for reliable real estate market analysis for an informed and handsome investment decision making. The study examines the opportunities and challenges of using artificial intelligence (AI) approaches to analyse quantitative / data on the Africa real estate market. A systematic review of the literature was conducted. The study found some opportunities characterised by the usage of AI tools to include a reliable forecasting model, good predictive power, ability to perform complex tasks and analyses of a large number of variables. The major constraints of AI tools application in the study area include knowledge gap, weak political will, unstable economic conditions, low level of market transparency, maturity, openness and lack of availability and reliable data and research on the Africa real estate market analysis (AREMA). The study concluded that the Africa real estate market is yet to be ready for the adoption of hi-tech market analysis devices such as AI and machine learning (ML) due to lack of availability and reliable data. However, the study suggests a strong collaboration and cooperation between the public and private sectors to create enabling real estate business environment that priorities technology infrastructure, data-driven property market and reliable data bank to achieve a sustainable Africa real estate market analysis for an informed investment decision.
... This result is quite different from the position of some researchers such as Azmi, Nawawi, Ab Latif, and Ling (2013) and Abidoye and Chan (2017b) that reported that the adoption of AI property valuation methods would transform the Malaysian and Nigerian property valuation practice, respectively, and impliedly the national economy. Also, researchers have argued that AI valuation methods could help valuers to save time, cost and reduce inaccuracy (Abidoye & Chan, 2017a;Chaphalkar & Sayali, 2013;Morano, Tajani, & Torre, 2015;Taffese, 2006), which is beneficial to the property industry (Abidoye, 2017). ...
Embracing technological advancement in the property valuation practice is unavoidable. However, studies show that valuers largely still adopt traditional methods of valuation. Hence, this study investigates the barriers, drivers, and prospects of the adoption of artificial intelligence (AI) valuation methods in practice. An online questionnaire survey was conducted on valuers practising in Australia. Their opinions about the topic were collected and analysed using frequency distribution and mean score ranking to establish the most significant factors. According to the valuers, the most important advantage of AI valuation methods is that they will help to reduce the cost of valuations. It was also found that the professional bodies that regulate the property valuation practice are the major driver of the adoption of AI valuation methods. The valuers expressed that AI valuation methods may produce accurate estimates. The valuers confirmed that the main prospect of adopting the AI valuation methods in practice is that it could transform the property valuation industry. It is evident that all the property valuation stakeholders should invest efforts in promoting the adoption of AI valuation methods in practice to bridge the gap between theory and practice. This will help reposition the property valuation profession.
... Machine learning based models can be predictive to carryout prediction or descriptive to discover knowledge from data, or both without assuming a predetermined equation as a model [90,95,96]. It is one of the foremost outstanding technological developments in recent history and used in various disciplines, e.g., computational finance [97][98][99], image and speech processing [100][101][102], property valuation [103][104][105], computational biology [106][107][108], and energy production [109][110][111]. Deep learning is a subset of machine learning that uses multiple layers to progressively extract features from the raw input data [112]. ...
Full-text available
In this work, technological feasibility of autonomous corrosion assessment of reinforced concrete structures is studied. Corrosion of reinforcement bars (rebar), induced by carbonation or chloride penetration, is one of the leading causes for deterioration of concrete structures throughout the globe. Continuous nondestructive in-service monitoring of carbonation through pH and chloride ion (Cl−) concentration in concrete is indispensable for early detection of corrosion and making appropriate decisions, which ultimately make the lifecycle management of RC structures optimal from resources and safety perspectives. Critical state-of-the-art review of pH and Cl − sensors revealed that the majority of the sensors have high sensitivity, reliability, and stability in concrete environment, though the experiments were carried out for relatively short periods. Among the reviewed works, only three attempted to monitor Cl− wirelessly, albeit over a very short range. As part of the feasibility study, this work recommends the use of internet of things (IoT) and machine learning for autonomous corrosion condition assessment of RC structures.
... Likewise, ANN widely applied in other industries like in real estate management [236] security [237], in business [238]. ...
Full-text available
The era of artificial neural network (ANN) began with a simplified application in many fields and remarkable success in pattern recognition (PR) even in manufacturing industries. Although significant progress achieved and surveyed in addressing ANN application to PR challenges, nevertheless, some problems are yet to be resolved like whimsical orientation (the unknown path that cannot be accurately calculated due to its directional position). Other problem includes; object classification, location, scaling, neurons behavior analysis in hidden layers, rule, and template matching. Also, the lack of extant literature on the issues associated with ANN application to PR seems to slow down research focus and progress in the field. Hence, there is a need for state-of-the-art in neural networks application to PR to urgently address the above-highlights problems for more successes. The study furnishes readers with a clearer understanding of the current, and new trend in ANN models that effectively addresses PR challenges to enable research focus and topics. Similarly, the comprehensive review reveals the diverse areas of the success of ANN models and their application to PR. In evaluating the performance of ANN models, some statistical indicators for measuring the performance of the ANN model in many studies were adopted. Such as the use of mean absolute percentage error (MAPE), mean absolute error (MAE), root mean squared error (RMSE), and variance of absolute percentage error (VAPE). The result shows that the current ANN models such as GAN, SAE, DBN, RBM, RNN, RBFN, PNN, CNN, SLP, MLP, MLNN, Reservoir computing, and Transformer models are performing excellently in their application to PR tasks. Therefore, the study recommends the research focus on current models and the development of new models concurrently for more successes in the field.
In the current scenario circular economy has become an important aspect of organization because of social, economic and environmental responsibility. The circular economy approaches are useful in resource conservation and based on the reducing, recycling and reuse philosophy. Further, industry 4.0 is coupled with the advanced technology such as Artificial Intelligence, Cybersecurity, Cyber-Physical-Systems, Internet of Things, Cloud Computing and Big Data. These technologies have the capability to enable environmentally sustainable manufacturing. The implementation of industry 4.0 in the circular economy viewpoint may provide data transparency, reliability, real- time information, and maximize the circularity of resources that improves the reverse logistic performance. Therefore, the present study seeks to investigate the several industry 4.0 and circular economy circularity decisions which provide a methodology for the selection of the most optimum circularity alternative using Graph Theory and Matrix Approach. Based on the comprehensive literature review and experts opinions, the study identified and prioritized the circularity alternatives of industry 4.0 and circular economy to improve the triple bottleneck of sustainability within the organization. A case of an Indian barrel manufacturing organization is considered for the illustration of this methodology. The value of a permanent function is calculated for each alternative using C++ language and alternatives are prioritized. The results show that “Digitalization” is the most optimum attribute while “Reduce” is the most optimum alternative to the circularity. The findings of the study will be valuable to industrialists and practitioners in strategic decision-making for the implementation of circularity in the manufacturing process.
Full-text available
Purpose The assessment of the Real Estate (RE) prices depends on multiple factors that traditional evaluation methods often struggle to fully understand. Housing prices, in particular, are the foundations for a better knowledge of the Built Environment and its characteristics. Recently, Machine Learning (ML) techniques, which are a subset of Artificial Intelligence, are gaining momentum in solving complex, non-linear problems like house price forecasting. Hence, this study deployed three popular ML techniques to predict dwelling prices in two cities in Italy. Design/methodology/approach An extensive dataset about house prices is collected through API protocol in two cities in North Italy, namely Brescia and Varese. This data is used to train and test three most popular ML models, i.e. ElasticNet, XGBoost and Artificial Neural Network, in order to predict house prices with six different features. Findings The models' performance was evaluated using the Mean Absolute Error (MAE) score. The results showed that the artificial neural network performed better than the others in predicting house prices, with a MAE 5% lower than the second-best model (which was the XGBoost). Research limitations/implications All the models had an accuracy drop in forecasting the most expensive cases, probably due to a lack of data. Practical implications The accessibility and easiness of the proposed model will allow future users to predict house prices with different datasets. Alternatively, further research may implement a different model using neural networks, knowing that they work better for this kind of task. Originality/value To date, this is the first comparison of the three most popular ML models that are usually employed when predicting house prices.
Full-text available
Accurate house price estimation is of great importance not only to various real estates stakeholders such as house owners, buyers, investors, and agents but also to banking and insurance sectors among others. This estimation is a very challenging and complex task as it involves a variety of attributes. Among all attributes, location is one of the most influential attributes affecting house price. The present study has incorporated selected structural attributes and relative location through spatial attributes (geographical, infrastructural, and neighborhood) for modeling the influence of location on house prices. The hedonic model, a traditional method for estimating house price has been criticized due to nonlinearity, multicollinearity, and heteroskedasticity problems. Unlike the hedonic model, the Artificial Neural Network (ANN) permits nonlinear relationships and also tries to solve the problem of multicollinearity. The present study aims to examine the application of ANN in estimating accurate house prices and comparing the results with the Ordinary Least Squares (OLS) based hedonic model. The Gotri area located in the western part of Vadodara city, India is considered as a case study. TensorFlow (a python environment) of Google is used to implement the ANN model. In ANN model building, a three-layer network with a single hidden layer and RELU (REctified Linear Unit) activation function is adopted. The estimation performance has been evaluated by employing Root Mean Squared Error and Mean Absolute Percentage Error. According to results, ANN was found better when compared to OLS in terms of both the performance measures. This paper suggests that the ANN estimator could be a complement to the OLS based linear regression method.
Full-text available
Introduces the theory of artificial neural networks (ANN). Discusses its application to the valuation of residential apartments. Compares the performance of the back propagation neural network (BP) model in estimating sale prices of apartments against the traditional multiple regression analysis (MRA) model. Concludes that the neural network model is an easy-to-use, black-box alternative to the MRA model.
Full-text available
Introductory text concentrating on experience factory
Hybrid systems as the next generation of intelligent applications within the field of mass appraisal and valuation are investigated. Motivated by the obvious limitations of paradigms that are being used in isolation or as stand-alone techniques such as multiple regression analysis, artificial neural networks and expert systems. Clearly, there are distinct advantages in integrating two or more information processing systems that would address some of the discrete problems of individual techniques. Examines first, the strategic development of mass appraisal approaches which have traditionally been based on “stand-alone” techniques; second, the potential application of an intelligent hybrid system. Highlights possible solutions by investigating various hybrid systems that may be developed incorporating a nearest neighbour algorithm (k-NN). The enhancements are aimed at two major deficiencies in traditional distance metrics; user dependence for attribute weights and biases in the distance metric towards matching categorical variables in the retrieval of neighbours. Solutions include statistical techniques: mean, coefficient of variation and significant mean. Data mining paradigms based on a loosely coupled neural network or alternatively a tight coupling with genetic algorithms are used to discover attribute weights. The hybrid architectures developed are applied to a property data set and their performance measured based on their predictive value as well as perspicuity. Concludes by considering the application and the relevance of these techniques within the field of computer assisted mass appraisal.
The potential application of data mining techniques in the extraction of information from property data sets is discussed. Particular interest is focused upon neural networks in the valuation of residential property with an evaluation of their ability to predict. Model testing infers a wide variation in the range of outputs with best results for stratified market subsets, using postal code as a locational delimiter. The paper questions whether predicted outcomes are within the range of valuation acceptability and examines issues relating to potential biasing and repeatability of results.
In recent years, neural networks are widely being used in areas where conventional statistical methods were used. This paper compares regression and neural networks on a real life data and two simulated examples. The results show that regression is much better than neural networks for skewed data. General guidelines are suggested to improve the performance of neural networks for skewed data.
Pt. I. Fundamentals of hybrid intelligent systems and agents -- 1. Introduction -- 2. Basics of hybrid intelligent systems -- 3. Basics of agents and multi-agent systems -- Pt. II. Methodology and framework -- 4. Agent-oriented methodologies -- 5. Agent-based framework for hybrid intelligent systems --6. Matchmaking in middle agents -- Pt. III. Application systems -- 7. Agent-based hybrid intelligent system for financial investment planning -- 8. Agent-based hybrid intelligent system for data mining -- Pt. IV. Concluding remarks -- 9. The less the more -- App. Sample source codes of the agent-based financial planning system
This paper compares the predictive performance of artificial neural networks (ANN) and multiple regression analysis (MRA) for single family housing sales. Multiple comparisons are made between the two data models in which the data sample size is varied, the funcional specifications is varied, and the temporal prediction is varied. We conclude that ANN performs better than MRA when a moderate to large data sample size is used. For our application, this "moderate to large data sample size" varied from 13% to 39% of the total data sample (506 to 1506 observations out of 3906 total observations). Our results give a plausible explanation why previous papers have obtained varied results when comparing MRA and ANN predictive performance for housing values.
This research applies neural network (NN) technology to real estate appraisal and compares the performance of two NN models in estimating the sales price of residential properties with a traditional multiple regression model. The study is based on 288 sales of homes in Fort Collins, Colorado. Results do not support previous findings that NNs are a superior tool for appraisal analysis. Furthermore, significant problems were encountered with the NN models: inconsistent results between packages, inconsistent results between runs of the same package, and long run times. Any appraiser who plans on using this new technology would do so with caution.
Predicting Housing Value: [16 000 The Development of a [17 P. 2000. Using Expert Systems and Artificial [18 tegrating Artificial Neural [19 lligence and the [20 eural [21 d A Comparison of Multiple Regression Analysis and Artificial Neural Networks
  • N Nguyen
Nguyen, N. et al. A. 2001. Predicting Housing Value: [16 000. The Development of a [17 P. 2000. Using Expert Systems and Artificial [18 tegrating Artificial Neural [19 lligence and the [20 eural [21 d A Comparison of Multiple Regression Analysis and Artificial Neural Networks. Journal of Real Estate Research 22(3): 313-335.