(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 5, 2010
Predicting Students' Academic Performance UsingArtificial Neural Networks: A Case Study
Ghaleb A. El-Refae
Faculty of Economics and Admin. SciencesAl-Zaytoonah University of JordanAmman, Jordanghalebalrefae@yahoo.com
Qeethara Kadhim Al-Shayea
Faculty of Economics and Admin. Sciences, MIS Dep.Al-Zaytoonah University of JordanAmman, Jordankit_alshayeh@yahoo.com
Abstract
—
Predicting students’ academic performance iscritical for universities because strategic programs can beplanned in improving or maintaining students’ performance.The goal of this study is to predict the factors affecting theuniversity students' performance using Artificial NeuralNetworks (ANN) model. Various factors that may likelyinfluence the performance of a student were identified.Generalized Regression Neural Network (GRNN) is used topredict the university students' performance.
It is noticed asignificant improvement in the prediction made byGRNN due to its generalization property
. The mostimportant predictor variable influencing performance isconsistently having the largest regression. Results showed thatsecondary school performance which is measured by scores insecondary school certificate examination, measured in apercentage form having the largest regression value.
Keywords-component; regression, stdudent performance; Artificial neural networks; general regression network
I.
I
NTRODUCTION
The prediction and explanation of academic performanceand the investigation of the factors relating to the academicsuccess and persistence of students are topics of utmostimportance in higher education [1].McKenzie and Schweitzer [2] presented a study that wasa prospective investigation of the academic, psychosocial,cognitive, and demographic predictors of academicperformance of first year Australian university students.Alfan [3] determined the undergraduate students'performance in the Faculty of Business and Accountancy,University of Malaya and the factors influencing theperformance of the undergraduate students. The result of thestudy shows that the predictor variables do explain thevariance in the students' final cumulative grade pointaverage. In addition, it was found that knowledge prior toentering the university such as economics, mathematics andaccounting is crucial in assisting the students in undertakingthe courses in both business and accounting program. Thestudy also found that female students perform better thanmale students; whilst Chinese students perform better thanMalay and Indian students.Su [4] evaluated the performance of university studentswho learned science texts by using, informationcommunication technologies including animation, staticfigures, power point, and e-plus software. The resultsincluded the computation of the F-ratio, p-values, andCohen’s effect-sizes of attitudes toward science and learningscience in relation to the student’s gender, attendance of computer orientation classes, use of computer-multimedia,disposition toward computers, and majors.Al-Tamimi and Al-Shayeb [5] investigated some factorsaffecting student performance in the fundamentals of financial management course at United Arab EmiratesUniversity.Ibrahim and Rusli [6] developed three predictive modelsusing SAS Enterprise Miner that are, artificial neuralnetwork, decision tree and linear regression. The result of this study showed that all of the three models produce morethan 80% accuracy. It also showed that artificial neuralnetwork outperforms the other two models.Oladokun, Adebanjo and Charles-Owaba [7] presentedan artificial neural network model for predicting the likelyperformance of a candidate being considered for admissioninto the university was developed and tested.II.
A
RTIFICIAL
N
EURAL
N
ETWORKS
An artificial neural network (ANN) is a computationalmodel that attempts to account for the parallel nature of thehuman brain. An (ANN) is a network of highlyinterconnecting processing elements (neurons) operating inparallel. These elements are inspired by biological nervoussystems. As in nature, the connections between elementslargely determine the network function. A subgroup of processing element is called a layer in the network. The firstlayer is the input layer and the last layer is the output layer.Between the input and output layer, there may be additionallayer(s) of units, called hidden layer(s). Fig. 1 represents thetypical neural network. You can train a neural network toperform a particular function by adjusting the values of theconnections (weights) between elements.
Figure 1. A typical neural network
97http://sites.google.com/site/ijcsis/ISSN 1947-5500
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 5, 2010
For the researcher and the financial analyst, the mainadvantage of ANNs is that there is no need to specify thefunctional relation between variables. Since they areconnectionist-learning machines, the knowledge is directlyimbedded in a set of weights through the linking arcs amongthe processing nodes. In order to train a neural network properly one needs a large set of representative 'goodquality’ examples. In the case of bankruptcy problems, theresearcher should be cautious when drawing conclusionsfrom neural networks trained with only one or two hundredcases, as observed in most previous studies [8].
A.
Generalized Regression Neural Network
The GRNN was applied to solve a variety of problemslike prediction, control, plant process modeling or generalmapping problems [9].General regression neural network Specht [10], Nadaraya [11] and Watson [12], does not require an iterativetraining procedure as in back-propagation method.The GRNN is used for estimation of continuousvariables, as in standard regression techniques. It is relatedto the radial basis function network and is based on astandard statistical technique called kernel regression. Bydefinition, the regression of a dependent variable y on anindependent x estimates the most probable value for y, givenx and a training set. The regression method will produce theestimated value of y, which minimizes the mean-squarederror. GRNN is a method for estimating the joint probabilitydensity function (pdf) of x and y, given only a training set.Because the pdf is derived from the data with nopreconceptions about its form, the system is perfectlygeneral. Furthermore, it is consistent; that is, as the trainingset size becomes large, the estimation error approaches zero,with only mild restrictions on the function. In GRNN,instead of training the weights, one simply assigns to wij thetarget value directly from the training set associated withinput training vector i and component j of its correspondingoutput vector [13]. GRNN architecture is given in Fig. 2.GRNN is based on the following formula [14]:(1)where
y
is the output of the estimator,
x
is the estimatorinput vector,
E[y|x]
is the expected output value, given theinput vector
x
and
f(x,y)
is the joint probability densityfunction (pdf) of
x
and
y
.The function value is estimated optimally as follows:(2)where
w
ij
= the target output corresponding to input trainingvector
x
i
,
22
.2
spread Di
i
eh
−
=
, the output of the hidden layer neuron,
( ) ( )
iT ii
u xu x D
−−=
2
, the squared distance between theinput vector
x
and the training vector
u
,
x
= the input vector,
u
i
=training vector i, the center of neuron i, spread=a constantcontrolling the size of the receptive region.
Figure 2. Generalized Regression Neural Network (GRNN) Architecture
III.
E
XPERIMENTAL
R
ESULTS
A.
Data
This study was conducted at the faculty of Economicsand Administrative Sciences, Al-Zaytoonah University of Jordan in Hashemite Kingdom of Jordan. Our sampleconsists of 208 students belonging to accountingdepartment. The information for this study has beenobtained from the register office at Al-Zaytoonah Universityof Jordan and which are maintained on a computerizeddatabase.The Cumulative Grade Average Point (CGPA) is used asan indicator to measure the performance of the universitystudents'.The students' overall performance was hypothesized tobe a function of the following factors: (1) Secondary schoolperformance is measured by scores in secondary schoolcertificate examination, measured in a percentage form (2)type of secondary school branch, (3) gender, and (4)boarding or non boarding student.
B.
Results Analysis
A generalized regression neural network (GRNN) with
a
radial basis layer and a special linear layer and linear outputneurons was created using the neural network toolbox fromMatlab 7.9 as shown in Fig. 2. Generalized regression neuralnetworks are a kind of radial basis network that is often usedfor function approximation.The first layer has as many neurons as there are input/ target vectors. Each neuron's weighted input is the distancebetween the input vector and its weight vector. Eachneuron's net input is the product of its weighted input withits bias. Each neuron's output is its net input passed throughradial basis transfer function. Radial basis transfer functionis a neural transfer function which calculates a layer's outputfrom its net input. If a neuron's weight vector is equal to theinput vector (transposed), its weighted input will be 0, its net
[ ]
∫∫
∞∞−∞∞−
=
dy y x f
dy y x f y
x y E
).,().,(.
|
∑∑
−=
=
niiniiji j
hwh y
11
.
98http://sites.google.com/site/ijcsis/ISSN 1947-5500
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 5, 2010
input will be 0, and its output will be 1. The second layeralso has as many neurons as input/target vectors.We used a spread slightly lower than the distancebetween input values, in order, to get a function that fitsindividual data points fairly closely. A smaller spread wouldfit data better but be less smooth.
Figure 3. A generalized regression neural network (GRNN)
The GRNN with cumulative grade point average CGPA as atarget and secondary school performances that is measuredby scores in secondary school certificate examination,measured in a percentage form as input was been created.Then simulate the network with 208 inputs. The network outputs after simulation. The spread value was chosen 0.2.The percent correctly predicted in the simulation sample isapproximately 76 percent as shown in Fig. 4.
Figure 4.
The GRNN with cumulative grade point average CGPAas a target and type of secondary school branch as input wasbeen created. The spread value was chosen 0.6. The percentcorrectly predicted in the simulation sample isapproximately 36 percent as shown in Fig. 5.
Figure 5.
The GRNN with cumulative grade point average CGPAas a target and gender as input was been created. The spreadvalue was chosen 0.5. The percent correctly predicted in thesimulation sample is approximately 27 percent as shown inFig. 6.
Figure 6.
The GRNN with cumulative grade point average CGPAas a target and boarding or non boarding as input was beencreated. The spread value was chosen 0.5. The percentcorrectly predicted in the simulation sample isapproximately 20 percent as shown in Fig. 7.
Figure 7.Figure 8.
99http://sites.google.com/site/ijcsis/ISSN 1947-5500
Search History:
Result 00 of 00
00 results for result for
p.
Predicting Students' Academic Performance Using Artificial Neural Networks: A Case Study
Predicting students’ academic performance is critical for universities because strategic programs can be planned in improving or maintaining students’ performance. The goal of this study is to predict the factors affecting the university students' performance using…
(More)
1.7K
Reads
17
Readcasts
1
Embed Views
Recommended
More From This User
Notes
Load more
