ARID ZONE JOURNAL OF ENGINEERING, TECHNOLOGY & ENVIRONMENT
AZOJETE June 2020. Vol. 16(2):243-254

Published by the Faculty of Engineering, University of Maiduguri, Maiduguri, Nigeria.
Print ISSN: 1596-2490, Electronic ISSN: 2545-5818

www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 243

ORIGINAL RESEARCH ARTICLE

COMPUTATIONAL MODEL OF ARTIFICIAL NEURAL NETWORKS AND ITS
APPLICATIONS IN DATA MINING

J. Abah1*, E. S. Alu2, A. E. Chahari3
(1Department of Computer Engineering, University of Maiduguri, Maiduguri, Nigeria
2Department of Agricultural Economics and Extension, Nasarawa State University,

Keffi, Shabu-Lafia Campus, Lafia, Nigeria
3Department of General Studies Education, School of General Education, Federal College of Education, Yola, Nigeria)

*Corresponding author’s email address: abah@unimaid.edu.ng

1.0 Introduction
As long as knowledge is another name of power, many organizations give much importance to
knowledge. When reaching this knowledge, they make use of the data in their databases. Data
is important since it enable them to learn from the past and to predict future trends and
behaviours (Larose, 2005). Today, most of the organizations use the data collected in their
databases when taking strategic decisions (Singh and Chauhan 2009). The process of using the
data to reach this knowledge consists of two steps as to collecting the data and analysing the
data. In the beginning, organizations are faced with difficulties when collecting the data so they
have not enough data in order to make suitable analysis (Kartalopoulos and Stamatios, 1996). In
the long run, with the rapid computerization, organizations are able to store huge amount of
data easily. But at this time, they are faced with another problem when analysing and

ARTICLE INFORMATION ABSTRACT

Data remain a very important ingredient required by any organization
to make informed decision as it affects operations. Companies have
been collecting data from various sources over the decades bringing
about a very large volume of data warehouse. Unfortunately, most
organizations build databases which are redundant and never used for
any meaningful thing. While few companies use the data collected in
their databases when taking strategic decisions others barely do same.
However, for an organization to immensely derive benefits from the
massive data warehouse, there is the need for an effective and efficient
means of analysing the data with a view to extracting meaningful
knowledge that is sufficient to achieve organizational goal. To achieve
this, Artificial Neural Network (ANN) technique through the concept
known as data mining is presented. The paper reviewed artificial neural
network technique for data mining, examines the computational model
behind this technique and analysed its use and application as a
predicting or forecasting tool. Results shows that ANN’ has capability in
data management, analysis and able to provide desirable knowledge for
management decision making processes. It is therefore recommended
that data mining tools like ANN and others be applied to organization’s
databases which hitherto have not been minned in order to provide
management with intelligence for decision making.

© 2020 Faculty of Engineering, University of Maiduguri, Nigeria. All rights reserved.

Submitted 11 July, 2017
Revised 15 October, 2019
Accepted 04 April, 2020

Keywords:
Artificial Neural Network
Databases
Data Mining
Information
Organizations

mailto:abah@unimaid.edu.ng1
http://www.azojete.com.ng


Abah et al: Computational Model of Artificial Neural Networks and its Applications in Data Mining.
AZOJETE,16(2):243-254. ISSN 1596-2490; e-ISSN 2545-5818, www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 244

interpreting such large datasets. Traditional methods like statistical techniques or data
management tools are not sufficient anymore (Berry and Linoff, 2000).

In order to manage this problem, the technique called Data Mining (DM) has been discovered
to address this problem. According to Singh and Chauhan (2009) show that data mining is a
useful and powerful technology that supports organizations to derive strategic information in
their databases. It has been defined as: “The process of exploration and analysis by automatic or
semi-automatic means of large quantities of data in order to discover meaningful patterns and
rules” (Berry and Linoff, 1997; Berry and Linoff, 2000). The expression “meaningful patterns and
rules” implies the followings; easily understood by humans, valid on new data, potentially useful
and novel. Validating a hypothesis that the user wants to prove can also be accepted as a
meaningful patterns and rules (Berry and Linoff, 2000). In sum, it is essential to derive patterns
and rules that help us to reach strategic and unimagined information in data mining.

Few organizations use the data collected in their databases when taking strategic decisions. This
has made most organization’s databases redundant. The process of using accumulated data to
reach knowledge consists of two steps; collecting the data and analyzing the data. The problem
associated with the process of achieving these steps are first, most organizations are faced with
difficulties when collecting the data so, they have not enough data in order to make suitable
analysis and second, the technique of analysing the collected data in order to derive meaning or
knowledge from it is often not sufficient and efficient to achieve the goal (Cilimkoviv, 2015).
In order to solve the stated problems, we explore ANN as a data mining technique to bring to
bear, it’s capability in data management and analysis and its ability to provide desirable
knowledge for informed management decision making processes. Hence, we present ANN as a
useful and powerful data mining technique that can support organizations to derive strategic
information from their databases. The rest of this paper is organized as follows: Section two
provides details about ANN; concept, models and topologies. Section three provides
experimental procedure of ANN technique, section four gives the discussion of results and
finally, section five gives the summary, conclusion and recommendation.

2. Artificial Neural Networks
An artificial neural network (ANN), just called a "neural network" (NN), is a mathematical model
or computational model based on biological neural networks (Berry and Linoff, 1997). In other
words, ANN is an emulation of biological neural system. It consists of an interconnected group
of artificial neurons and processes information using a connectionist approach to computation.
In most cases an ANN is an adaptive system that changes its structure based on external or
internal information that flows through the network during the learning phase (Singh and
Chauhan, 2009; Zurada, 1992).

2.1 Concept of Neural Network
Neural networks are very sophisticated modeling techniques capable of modelling extremely
complex functions. Neural networks learn by example. The neural network user gathers
representative data, and then invokes training algorithms to automatically learn the structure of
the data. It is a model for classification and prediction (continuous value) of an outcome variable
(Beale and Jackson, 1990).

http://www.azojete.com.ng


Arid Zone Journal of Engineering, Technology and Environment, June, 2020; Vol. 16(2) 243-254. ISSN 1596-2490;
e-ISSN 2545-5818; www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 245

2.2 The Basic Artificial model
The artificial neuron was developed in an effort to model the human neuron (Kartalopoulos and
Stamatios, 1996; Haykin and Simon, 1994). Inputs enter the neuron and are multiplied by their
respective synaptic weights. They are then summed and processed by an activation function.
Examples of activations functions includes the logistic or sigmoid function (1).

f(x) = 1/ (1+exp (-x)) (1)
while the second is the Gaussian function (2)

f(x) = exp (-x2). (2)
Where the variable x represents the input vector in both equations.

Figure 1: Artificial Neural Network Model

2.3 Neural Network Topologies
Feed Forward Neural Network: The feedforward neural network was the first and arguably
simplest type of artificial neural network devised (Singh and Chauhan, 2009). In this network, the
information moves in only one direction, forward, from the input nodes, through the hidden
nodes (if any) and to the output nodes. There are no cycles or loops in the network. The data
processing can extend over multiple (layers of) units, but no feedback connections are present,
that is, connections extending from outputs of units to inputs of units in the same layer or
previous layers (Zurada, 1992).
Recurrent network: Recurrent neural networks that do contain feedback connections. Contrary
to feedforward networks, recurrent neural networks (RNs) are models with bi-directional data
flow. While a feedforward network propagates data linearly from input to output, RNs also
propagate data from later processing stages to earlier stages (Singh and Chauhan, 2009).

2.4 Data Mining
Data mining is a set of techniques and procedures that can be developed from various data
sources such as data warehouses or relational databases (Villanueva et al., 2018). Flat files
without formats are made from this predictive analysis using statistical study techniques to
predict or anticipate statistical measures of certainty based on existing facts. Data mining is seen
as the evolution of information technology (Jiawei and Kamber, 2001), this is largely supported
by the growth of the internet. Data mining is a multidisciplinary field that allows relevant
information to be obtained from large amounts of data at the confluence among other
disciplines namely: artificial intelligence, statistics, databases and information science. Figure 2
shows data mining taxonomy.

-
f

Weighted sumInput vector x

Output y

Activationfunction

Bias

Weight vector w



w0j

w1j

wnj

x0

x1

xn

file:///C:/Users/user/Downloads/azojete143/www.azojete.com.ng


Abah et al: Computational Model of Artificial Neural Networks and its Applications in Data Mining.
AZOJETE,16(2):243-254. ISSN 1596-2490; e-ISSN 2545-5818, www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 246

Figure 2: Data Mining Taxonomy (Maimon and Rokach, 2010)

2.4.1 Phases of Data Mining Task
The data mining tasks are implemented with the goal of discovering patterns of relevant and
interesting information in large volumes. This is done with the development of four phases
according to Villanueva et al. (2018), which are usually:
Filtering data;
Selection of variables;
Extracting knowledge and
Interpretation and evaluation.
In general, all techniques have been proved in educational settings according to Siemens and
Baker (2012), and different case studies have been developed to evaluate the performance of
different techniques and to meet the main goals of data mining.

2.4.2 Neural Networks in Data Mining
In more practical terms neural networks are non-linear statistical data modelling tools. They can
be used to model complex relationships between inputs and outputs or to find patterns in data
(Singh and Chauhan, 2009). Using neural networks as a tool, data warehousing firms are
harvesting information from datasets in the process known as data mining. The difference
between these data warehouses and ordinary databases is that there is actual manipulation and
cross-fertilization of the data helping users make more informed decisions. Neural networks
essentially comprise three pieces (Zurada, 1992): The architecture or model; the learning
algorithm; and the activation functions. Neural networks are programmed or “trained” to store,
recognize, and associatively retrieve patterns or database entries; to solve combinatorial
optimization problems; to filter noise from measurement data; to control ill-defined problems;
in summary, to estimate sampled functions when we do not know the form of the functions.” It
is precisely these two abilities (pattern recognition and function estimation) which make artificial
neural networks (ANN) so prevalent a utility in data mining.

According to Villanueva, Moreno and Salinas (2018), the first known work on Educational Data
Mining titled "Explaining student grades Predicted by a neural network" (Gedeon and Turner,
1993). In this work, the researchers used neural networks were used to predict the final grades
of students. Later, in 1994, Fausett and Elwasif published "Predicting performance from test

http://www.azojete.com.ng


Arid Zone Journal of Engineering, Technology and Environment, June, 2020; Vol. 16(2) 243-254. ISSN 1596-2490;
e-ISSN 2545-5818; www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 247

scores using back propagation and counter propagation" in which neural networks are used to
predict the student’s performance (Fausett and Elwasif, 1994).

"Predicting and analyzing secondary education placement-test scores: A data mining approach"
by Şen, Ucar and Delen (2012) relied on neural networks and decision trees to predict
outcomes of high school students in Turkey. Bayesian networks were used to work educational
situations, the first of them is "Predicting student's academic performance artificial using neural
network: A case study of an engineering course" (Oladokun et al., 2008) where the researchers
explain how to predict the performance a candidate might have if it is accepted in some
university courses.

Additionally, Sundar (2013) in his work "A Comparative Study for Predicting Students Academic
Performance using Bayesian Network Classifiers" made a comparison of obtained results using
Bayesian networks in predicting student performance; Likewise, Bhise et al.,(2013) published
"Importance of Data Mining in Higher Education System" their work is based on clustering to
help instructors improving student performance.

Today, educational and business information systems store large volume of data and its origin
can come from different sources, different formats and different granularity levels (Villanueva et
al., 2018). The problems of business and educational data mining must be analyzed particularly
due to their specific objective determines a singularity when it is solved by data mining
techniques. For instance, in the domain of education, Data Mining for education is so important
that in 2007 an international organization dedicated to researching this discipline was founded.
Such is the importance has prompted this research area that it is estimated that by the year
2022 all research related to education will involve analysis and data mining (Baker and Yacef,
2009).

2.5 Feedforward Neural Network
One of the simplest Feedforward Neural Networks (FFNN), such as in figure 3, consists of three
layers: an input layer, hidden layer and output layer. In each layer there are one or more
processing elements (PEs) (Singh and Chauhan, 2009). PEs is meant to simulate the neurons in
the brain and this is why they are often referred to as neurons or nodes.

x1

xn
Figure 3: Feedforward Neural Network

A PE receives inputs from either the outside world. The simplified process for training a FFNN is
as follows:
1. Input data is presented to the network and propagated through the network until it reaches
the output layer. This forward process produces a predicted output.

PE

PE

PE

file:///C:/Users/user/Downloads/azojete143/www.azojete.com.ng


Abah et al: Computational Model of Artificial Neural Networks and its Applications in Data Mining.
AZOJETE,16(2):243-254. ISSN 1596-2490; e-ISSN 2545-5818, www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 248

2. The predicted output is subtracted from the actual output and an error value for the networks
is calculated.
3. The neural network then uses supervised learning, which in most cases is back propagation,
to train the network. Back propagation is a learning algorithm for adjusting the weights. It starts
with the weights between the output layer PE’s and the last hidden layer PE’s and works
backwards through the network.
4. Once back propagation has finished, the forward process starts again, and this cycle is
continued until the error between predicted and actual outputs is minimized.

2.6 Artificial Neuron using Back Propagation Learning
The mathematical formulae used for Neural Network in Back propagation learning algorithm
according to Dhar et al. (1996) presented in Figure 4, was used.

Figure 4: Artificial Neuron using Back Propagation Learning

Figure 4 depicts a single artificial neuron and the actual computation that occurs in neural
network in a single epoch through the data.

2.7 Fitting a Network to Data
The equations in Figure 4 provides a good understanding on how neural network process data.
However, for the purpose of clarity and understanding, we illustrate the actual computation that
occurs in neural network in a single epoch through the data. Typically, neural network performs
exceptionally well with large dataset but for the purpose of this illustration a small dataset is
used as shown in Table 1. Refer to Figure 5 for the simulated neural network for the dataset
provided in Table 1.

Table 1: Sample Dataset
Observation Milk Score Sugar Score Price
1 0.2 0.9 0.65
2 0.1 0.1 0.34
3 0.2 0.4 0.42
4 0.2 0.5 0.49
5 0.4 0.5 0.72
6 0.3 0.8 0.83

http://www.azojete.com.ng


Arid Zone Journal of Engineering, Technology and Environment, June, 2020; Vol. 16(2) 243-254. ISSN 1596-2490;
e-ISSN 2545-5818; www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 249

Figure 5: Simulated Neural Network

Considering the first observation: for the input at the input layer; milk = 0.2 and sugar = 0.9, the
output of this layer is x1 = 0.2 and x2 = 0.9.
Hidden layer nodes take as input the output values from the input layer. The hidden layer in this
example consists of 3 nodes. Each of the nodes receives input from all the input nodes (Figure
4). To compute the output of a hidden layer node, the weighted sum of the inputs is computed
and then applies a certain function to it.

3. Experimental Procedure
The procedure describes the steps involved in the computational process. First we describe the
artificial neural network as follows;

3.1 The Artificial Neural Network Model
Let x1, x2,…, xp be the set of input values to a node, the weighted sum is computed thus;
From the general algorithm λj + i=1

p wij xi�
We have;

 


n

i iijjin xwby
1 (3)

where: b, wi,j,…,wnj are weights that are initially set randomly then adjusted as the network
“learns”. the variable b, also known as the “bias” of node j, is a constant that controls the level of
contribution of node j.
The next step is to apply a function called transfer function (that is, linear function, exponential
function and a logistics/sigmoidal function).
In this paper, the sigmoid function in (1) is used for its practical value from the fact that it has a
squashing effect on very small or very large value.

Output LayerHidden LayersInput Layers

Milk

SUGAR

W15

W23

W24

W25

W14

W13

W36

W46

W56

b3

b4

b5

1

2

3

4

5

6

b6

file:///C:/Users/user/Downloads/azojete143/www.azojete.com.ng


Abah et al: Computational Model of Artificial Neural Networks and its Applications in Data Mining.
AZOJETE,16(2):243-254. ISSN 1596-2490; e-ISSN 2545-5818, www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 250

Node6 =

Node6 =

Using the logistic function then, the output of node j in the hidden layer is computed from the
general formula:

   




n

i xwbiijjj n

i iijje
xwbgoutput

1
11

1)(
(4)

as follows:
g λj + i=1

p wij xi� = 1

1+ e− λj+ i=1
p wij xi�

(5)

where:
Wij is the weighted vector between i and j processing elements;
Xi is the input vector in i processing element;
p is the processing element
bj and λj are the bias in (4) and (5) respectively.

3.2 Initializing the Weights
At this stage of the procedure, the values of bj and wij are typically initialize to small (but
generally random) number in the range 0.00 ± 0.05. With this value range, we can then compute
the outputs of Node3, Node4 and Node5. Suppose the following are the initial weights for;
Node 3 are bnw3=b3=-0.3, w13=0.05 and w23=0.01
Node 4 are bnw4=b4= 0.2, w14=-0.01 and w24=0.03
Node 5 are bnw5=b5=0.05, w15=0.02 and w25=-0.01
Node 6 are bnw6=b6=-0.015, w36=0.01, w46=0.05 and w56=0.015
where: the input x1 and x2 are 0.2 and 0.9 respectively as seen in observation 1 in table 1.
Using the following values in (2), the output of node3, Node4, Node5 are computed as follows;

The output from node3, node4 and node5 are fed into the output layer (node6) to compute the
final output for the first observation in our sample data.

3.3 Training the Model
After initializing the weights, the next step is to train the model. Training the model means
estimating the weights bj (bias node) and wij (link between nodes) that lead to the best
predictive results. The process illustrated above for computing the neural network output for an

Node3 = = 0.4302

Node4 = = 0.5560

= 0.5112
Node5 =

= 0.5062

http://www.azojete.com.ng


Arid Zone Journal of Engineering, Technology and Environment, June, 2020; Vol. 16(2) 243-254. ISSN 1596-2490;
e-ISSN 2545-5818; www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 251

observation is repeated for all the observations in the training set. For each observation, the
model produces a prediction which is then compared with the actual response value. Their
difference is the error for the output node. This error obtained from the output node is use
iteratively to update the estimated weights. In particular, the error for the output node is
distributed across all the hidden nodes that led to it, so that each node is assigned
“responsibility" for part of the error. Each of these node-specific errors is then used for updating
the weights.

3.4 Back Propagation of Error
Back propagation is a process for using model errors to update weights (“learning”). As the
name implies, errors are computed from the output layer back to the hidden layers. The error
associated with output node is computed by the general formula;

Errj = Nj (1-Nj) (Dj-Nj) (6)
The above equation is similar to the ordinary definition of an error (Dj – Nj) multiply by a
correction factor. The weights are then updated as follows:

ibjnew = bjold + lErrj (7)
wijnew = wijold + lErrj (8)

where: l is a learning rate or weight decay parameter, a constant ranging typically between 0
and 1, which controls the amount of change in weights from one iteration to the other.
After the first iteration, the error associated with the output node is;
Err = 0.5062(1-0.5062) (0.65-0.5062) = 0.03594.
Hence the weight update for second iteration is computed as follows using l = 0.5
b3: -0.3 + 0.5 (0.03594) = -0.282
b4: 0.2 + 0.5 (0.03594) = 0.218
b5: 0.05 + 0.5 (0.03594) = 0.068
b6: -0.015 + 0.5 (0.03594) = 0.003
W13: 0.08 + 0.5 (0.03594) = 0.068
W23: 0.01 + 0.5 (0.03594) = 0.028
W14: -0.01 + 0.5 (0.03594) = 0.008
W24: 0.03 + 0.5 (0.03594) = 0.048
W15: 0.02 + 0.5 (0.03594) = 0.038
W25: -0.01 + 0.5 (0.03594) = 0.008
W36: 0.01 + 0.5 (0.03594) = 0.028
W46: 0.05 + 0.5 (0.03594) = 0.068
W56: 0.015 + 0.5 (0.03594) = 0.033

The adjusted weights are used on the second observation (i.e. x1=0.1 and x2=0.1) to computer
the second output of the model. This process is repeated for the entire training data (in this
case 6). Hence one sweep through the data, is called an epoch, consisting of six (6) iterations.
For the purpose of this simulation, the network is trained using 5 epochs so, there will be a total
of 30 iterations.

4. Results and Discussion

file:///C:/Users/user/Downloads/azojete143/www.azojete.com.ng


Abah et al: Computational Model of Artificial Neural Networks and its Applications in Data Mining.
AZOJETE,16(2):243-254. ISSN 1596-2490; e-ISSN 2545-5818, www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 252

Table 2 reveal the computation result of the adjusted weights, hidden layers output and the
final output for each epoch and the total sum of error associated with each epoch.

Table 2: Output Errors Generated for each of the Five (5) Epochs

From Table 2, we observed that the total sum of error decreases after every epoch but after a
while it begins to increase (for a large dataset). The point of minimum validation error is a good
indicator of the best number of epoch for training and the weights at that stage are most likely
to provide the best error rate in new data. However, it should be noted that the number of
observations in this simulation is too small for estimating the 13 weights, hence the reason for
the large error margin between the actual price and the predicted price (Table 3). The negative
margin signifies that the predicted value is more than the actual price. As the simulation runs
from the sart epoch (epoch 1) to the final epoch (epoch 5), the predicted value seems to
decrease hence, an increase in the margin. This is as a result of the learning process taking place
from one epoch to the next.

Table 3 gives the predicted values on the sample dataset. The predicted values were obtained
from the output values (output 6) of the last epoch (epoch 5).

http://www.azojete.com.ng


Arid Zone Journal of Engineering, Technology and Environment, June, 2020; Vol. 16(2) 243-254. ISSN 1596-2490;
e-ISSN 2545-5818; www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 253

Table 3: Predicted Values on a Sample Dataset

5. Conclusion
In this paper, we try to bring to the fore an overview of Artificial Neural Networks (ANN) as a
prediction tool. It is an attempt to build machine that will mimic brain activities and be able to
learn. ANN usually learns by examples. Basic ANN is composed of three layers, input, output
and hidden layer. Each layer can have number of nodes and nodes from input layer are
connected to the nodes from hidden layer. Nodes from hidden layer are connected to the
nodes from output layer. Those connections represent weights between nodes. It further
describes one of the most popular ANN algorithms; The Back Propagation (BP) Algorithm. The
idea behind BP algorithm is quite simple; output of ANN is evaluated against desired output. If
results are not satisfactory, connection (weights) between layers are modified and the process is
repeated again and again until error is small enough. Simple BP example is demonstrated in this
paper with ANN architecture also covered.

In conclusion, ANN is able to perform classification and even discover new trends or patterns in
data when supplied with enough samples. The dataset must be considerably large for ANN to
perform well. From the sample simulation, we were able to discover patterns or trend in the
price of the given commodities (milk and sugar in this case). As the simulation iterates from one
epoch to the next, the predicted value decreases to a given value before it increases again. The
minimum value attained before it starts to inrease again point to the best prediction. Similarly,
the observed decrease in the margin between stated price and predicted value shows a pattern
within the data which is not visible in the given dataset. This is an indication that in most cases
neural networks perform better than the traditional statistical techniques. Thus, neural networks
are becoming very popular with data mining practitioners, particularly in medical research,
finance and marketing. This is because they have proven their predictive power through
comparison with other statistical techniques using real data sets (Singh and Chauhan, 2009).
It is recommended therefore that organizations who want to make meaning from their huge
data repository should apply ANN models. This is not to say that ANN are perfect but due to
design problems neural systems need further research before they are widely accepted in
industry. As software companies develop more sophisticated models with user-friendly
interfaces, the attraction to artificial neural networks will continue to grow.

References

Baker, RSJ. and Yacef, K. 2009. The state of educational data mining in 2009: A Review and
Future visions. Journal of Educational Data Mining, 1(1): 3–17.

S/No. x1 x2 Price Predicted values Margin
(Price - Predicted value)

1 0.2 0.9 0.65 0.58 0.07
2 0.1 0.1 0.34 0.58 -0.24
3 0.2 0.4 0.42 0.57 -0.15
4 0.2 0.5 0.49 0.55 -0.06
5 0.4 0.5 0.72 0.55 0.17
6 0.3 0.8 0.83 0.56 0.27

file:///C:/Users/user/Downloads/azojete143/www.azojete.com.ng


Abah et al: Computational Model of Artificial Neural Networks and its Applications in Data Mining.
AZOJETE,16(2):243-254. ISSN 1596-2490; e-ISSN 2545-5818, www.azojete.com.ng

Corresponding author’s e-mail address: abah@unimaid.edu.ng 254

Beale, R. and Jackson, T. 1990. Neural Computing: An Introduction. Adam Hilger, Bristol,
England., pp.1-14.
Berry, MJA. and Linoff, GS. 1997. Data Mining Techniques. New York: Wiley and Sons., pp. 1-444.
Berry, MJA. and Linoff, GS. 2000. Mastering Data Mining: The art and Science of Customer
Relationship Management. New York: Wiley., pp. 1-512.
Bhise, RB., Thorat, SS. and Supekar, AK. 2013. Importance of Data Mining in Higher Education
System. IOSR Journal of Humanities and Social Science, 6(6): 18–21.
Dhar, V. and Stein R., 1996. Intelligent Decision Support Methods: The Science of Knowledge
Work, Prentice Hall Business Publishing. Upper Saddle River, New Jersey., pp. 11-94.
Fausett, L. and Elwasif, W. 1994. Predicting performance from test scores using back
propagation and counter propagation. Proceeding of 1994 IEEE International Conference on
Neural Networks (ICNN’94), 5: 3398–3402.
Gedeon, TD. and Turner, HS. 1993. Explaining Student Grades Predicted by a Neural Network. In
Proceeding of 1993 IEEE International Conference on Neural Networks (IJCNN-93-Nagoya,
Japan), 1: 609–612.
Jiawei, H. and Kamber, M. 2001. Data mining: concepts and techniques. San Francisco, CA, Ltd:
Morgan Kaufmann, pp. 377–385.
Larose, DT. 2005. Discovering knowledge in data: an introduction to data mining. Wiley-
InterScience., pp. 40-241.
Maimon, O. and Rokach, L. 2010. Data Mining and Knowledge Discovery Handbook 2nd Ed.
Springer New York Dordrecht Heidelberg London, 1-37.

Cilimkoviv, M., 2015. Neural Network and Back Propagation Algorithm, Institute of Technology
Blanchardstown, Blanchardstown Road North, Dublin, Ireland. 1-13.
Oladokun, VO., Adebanjo, AT. and Charles, B. 2008. Predicting Student’s Academic Performance
using Artificial Neural Network: A Case Study of an Engineering Course. The Pacific Journal of
Science and Technology, 9(1): 72–79.
Şen, B., Uçar, E. and Delen, D. 2012. Predicting and analyzing secondary education placement-
test scores: A data mining approach. Expert Systems with Applications, 39: 9468–9476.
Siemens, G. and Baker, RSJD. 2012. Learning analytics and educational data mining: Towards
Communication and Collaboration. LAK’12: Proceedings of the ACM 2nd International
Conference on Learning Analytics and Knowledge, Vancouver, BC, Canada. 252-254.
Sundar, PP. 2013. A Comparative Study for Predicting Students Academic Performance using
Bayesian Network Classifiers. IOSR Journal of Engineering, 3(2): 37–42.
Villanueva, A., Moreno, LG. and Salinas, MJ. 2018. Data mining techniques applied in educational
environments: Literature review. Digital Education Review, 33: 235-255.
Singh, Y. and Chauhan, A.S., 2009. Neural Network in Data Mining. Journal of Theoretical and
Applied Information Technology, 5(6): 36-42.
Zurada, JM. 1992. Introduction to Artificial Neural Systems. West Publishing Co., St. Paul, New
York, United States. pp. 1-764

http://www.azojete.com.ng