Empirical Study on Classification Algorithm For Evaluation of Students Academic Performance

IOSR Journal of Computer Engineering (IOSR-JCE)
e-ISSN: 2278-0661,p-ISSN: 2278-8727, Volume 17, Issue 6, Ver. II (Nov – Dec. 2015), PP 67-73
www.iosrjournals.org
DOI: 10.9790/0661-17626773 www.iosrjournals.org 67 | Page
Empirical Study on Classification Algorithm For Evaluation of
Students Academic Performance
Dr.S.Umamaheswari 1
, K.S.Divyaa 2
1
Associate Professor, School of IT and Science, Dr.G.R.Damodaran College of Science, Tamilnadu,
2
Research Scholar, Master of Philosophy, School of IT and Science, Dr.G.R.Damodaran College of Science,
Tamilnadu,
Abstract: Data mining techniques (DMT) are extensively used in educational field to find new hidden patterns
from student’s data. In recent years, the greatest issues that educational institutions are facing the unstable
expansion of educational data and to utilize this information data to progress the quality of managerial
decisions. Educational institutions are playing a prominent role in the public and also playing an essential role
for enlargement and progress of nation. The idea is predicting the paths of students, thus identifying the student
achievement. The data mining methods are very useful in predicting the educational database. Educational data
mining is concerns with improving techniques for determining knowledge from data which comes from the
educational database. However it has issue with accuracy of classification algorithms. To overcome this
problem the higher accuracy of the classification J48 algorithm is used. This work takes consideration with the
locality and the performance of the student in education in order to analyse the student achievement is high over
schooling or in graduation.
Keywords: Educational Data Mining, Performance Metrices.
I. Inroduction
Data mining (DM) is called as knowledge discovery in database (KDD), is known for its powerful role
in discovering hidden information from large volumes of data. Generally, data mining is the search for hidden
patterns that could be present in huge databases. Data mining is becoming gradually more important tool to
make over this data into information. Educational Data Mining (EDM) develops methods and applies techniques
from machine learning, statistics and data mining to analyse data collected during teaching and learning [1].
Educational Data Mining (EDM) is a growing field, concerned with developing methods for recognising the
unique characters of data that come from educational surroundings, and applying those methods to better
understand students, and helps in decision making. Educational data mining is an interesting research area which
extracts useful, previously unknown patterns from educational database for better understanding, improved
educational performance and assessment of the student learning process. Data resides from the Department of
School of Information Technology and science of the Dr.G.R.Damodaran College of Science. The Data
collection is done from the student database of B.Sc., (Computer Science) and B.Sc., (Information Technology)
for the past 3year’s i.e.2010-2012, 2011-2013, 2012-2015. The data is analysed in order to predict the
improvement over schooling or in graduation.
II. Evaluation Dataset
The data is collected in two different phases. Initially, the data collected at SSLC, HSC with school
name and UG level Mark/Percentage data. Data is collected from the Department of School of Information
Technology and science of the Dr.G.R.Damodaran College of Science. The Data collection is done from the
student database of B.Sc., (Computer Science) and B.Sc., (Information Technology) for the past 3year’s
i.e.2010-2012, 2011-2013, 2012-2015. The general attributes are student roll number, name, and gender, date of
birth, graduation year, address, phone number, location and city. The specific attributes are school name, school
location, student’s mark in school, college name, department, college location and student’s mark in college.
The algorithms are suggested to evaluate the performance of student in school academic and college academic.
The location details are such as urban school, urban home, rural school, urban college and rural home for
students. The specified dataset which provides more accurate analysis as well as prediction results based on the
clustering and classification algorithms. Secondarily the main parameters are considered.

Empirical Study On Classification Algorithm For Evaluation Of Students Academic Performance
Figure 1: The Student Data Set Description
In Data processing the data set used in this work contains graduate students information collected from
the college. The graduate student consists of 350 records and 15 attributes. Figure 1 presents the attributes and
their description that exists in the data set as taken from the source database. The selected attributes and
description are the selected for the analysis process.
 The data set contains some missing values in various attributes from 350 records; the records with
missing values are ignored from the data set since it doesn’t consider a large amount of data. The
number of records is reduced.
After applying pre-processing and preparation methods, analyse the data graphically and figure out the
grade of students using MAT Lab. The Data Set is shown in figure 2.
Figure:2 Sample Data
A. Train Data
The B.Sc., CS 2010-2013 batch dataset which contains 155 number of students. The training has been
done on the given dataset which shows the number of students with lowest percentage as 5. The number of
students with medium percentage is as 110. And the number of students with highest percentage is as 40. In this
dataset, it also considers all semesters such as semester 1 up to semester 6. And it provides low, medium and
highest percentages for all semesters. The random forest and J48 algorithm is used to train the specified dataset
based on the tree structured format.
B. Test Data
In this section we use the B.Sc., IT 2010-2013 batch dataset which contains counts of total students.
The testing has been done on the given dataset which shows the true positive, true negative, false positive and
false negative values. Then it is used to predict the model accuracy values for the specified datasets.
III. Classifiction Algorithm
A. Random Forest Classification Algorithm
The Figure 3 shows the information details about the student in the department of computer science of
the batch (2010-2013, 2011-2014, 2012-2015). The location of urban home students and urban college
Attribute Description
Stud_Rollno Student ID/ Roll Number
Stud_Name Name of the Student
Gender The gender of the student
Dob The date of birth of the student
Enrol_year The year of enrolment in the college
Gradu_year The year of graduation from the college
Home_Loca Location of the student home
Tel_no The telephone number of the student
HSC_Perc Percentage in the Higher Secondary Education
HSC_School School in which the Student have studied
HSC_Loca Location of the Higher Secondary Education
UG_Perc Percentage in the Under Graduation
UG_Loca Location of the UG College
S1,S2,S3…S6 Semester wise mark List
UG_Major Major of the Degree

constraints are shown the better performance of students. Total number of students is sixty and 59 student’s
study performance is increased and one student performance is reduced. The location of urban home students
and rural college constraints produced better performance. Total number of students is 17 and 16 student’s study
performance is increased and one student performance is decreased. The rural home students and urban college
conditions are produced good performance. Total number of students is 20 and 18 student’s educational
performance is increased and 2 student’s educational performance is decreased. The location of rural home
students and rural college conditions are produced good performance. Total number of students is 58 and 57
student’s educational performance is increased and 1 student’s educational performance is decreased.
Figure 3 : Random Forest Classification Algorithm
The B.Sc., CS (2010-2013, 2011-2014, 2012-2015) batches dataset which contains 155 number of
students. The training has been done on the given dataset which shows the number of students with lowest
percentage as 33. The number of students with medium percentage is as 82. And the number of students with
highest percentage is as 40. In this dataset, it also considers all semesters such as semester 1 up to semester 6.
And it provides low, medium and highest percentages for all semesters.
B. J48 Classification Algorithm
J48 builds decision trees from a set of training data in the same way as ID3, using the concept of
information entropy. The training data is a set S = s1,s2,... of already classified samples. Each sample si =
x1,x2,... is a vector where x1,x2,... represent attributes or features of the sample. The training data is augmented
with a vector C = c1,c2,... where c1,c2,... represent the class to which each sample belongs.
At each node of the tree, J48 chooses one attribute of the data that most effectively splits its set of
samples into subsets enriched in one class or the other. Its criterion is the normalized information gain
(difference in entropy) that results from choosing an attribute for splitting the data. The attribute with the highest
normalized information gain is chosen to make the decision. The J48 algorithm then recurs on the smaller sub
lists.
IV. Performance Evaluation
The Performance is evaluated for the existing and the proposed system. The analysis has been done for
the Random Forest and J48 classification Algorithms. In the existing system and the Proposed system the
accuracy, Precision, Recall, F-Measure is evaluated. From the experimental result, the scenario concludes that
the j48 algorithm yields greater accuracy performances. The higher performance are in terms of precision, recall
, accuracy and F-Measure metrics. From the Figure 4 describes that the existing and the proposed systems are
analyzed using Random Forest and J48 Classification Algorithm.

Figure 4 : Comparison Table
The evaluation is performed using the following performance metrices
 Precision
 Recall
 Accuracy
 F-Measure
To implement the proposed method and generate numerous results using mat lab tool in this
environment. The scenario has been selected educational dataset to discover the low, medium and high
performance of the students. In this section, the analysis has been done for existing and proposed research work
by using algorithms. The performance metrics are such as accuracy, precision, recall and f-measure values
which are evaluated by using random forest and J48 classification method. From the experimental result, the
conclusion decides that the proposed method provides higher performance results in terms of accuracy,
precision, recall and f-measure values.
Precision =
True positive
True positive +False positive
Precision can be seen as a measure of exactness or quality, whereas recall is a measure of completeness
or quantity. In simple terms, high precision means that an algorithm returned substantially more relevant results
than irrelevant. In a classification task, the precision for a class is the number of true positives (i.e. the number
of items correctly labeled as belonging to the positive class) divided by the total number of elements labeled as
belonging to the positive class (i.e. the sum of true positives and false positives, which are items incorrectly
labeled as belonging to the class).
Figure 5 : Precision

From the figure 5 can observe that the comparison of existing and proposed system in terms of
precision metric. In x axis we plot the methods and in y axis plot the precision values. In existing scenario, the
precision values are lower by using random forest algorithm. The precision value of existing scenario is 0.55 for
discover the student’s performance. In proposed system, the precision value is higher by using the J48
algorithm. The precision value of proposed scenario is 0.61 for discover the student’s performance. Thus it
shows that effective analysis is performed by using proposed algorithm. From the result, can conclude that
proposed system is superior in performance.
The calculation of the recall value is done as follows:
Recall =
True positive
True positive +False negative
The comparison graph is depicted as follows:
Recall is defined as the number of relevant documents retrieved by a search divided by the total
number of existing relevant documents, while precision is defined as the number of relevant
documents retrieved by a search divided by the total number of documents retrieved by that search. Recall in
this context is defined as the number of true positives divided by the total number of elements that actually
belong to the positive class (i.e. the sum of true positives and false negatives, which are items which were not
labeled as belonging to the positive class but should have been).
Figure 6 : Recall
From the figure 6 can observe that the comparison of existing and proposed system in terms of recall
metric. In x axis we plot the methods and in y axis we plot the recall values. In existing scenario, the recall
values are lower by using random forest algorithm. The recall value of existing scenario is 0.91 for discover the
student’s performance. In proposed system, the recall value is higher by using the J48 algorithm. The recall
value of proposed scenario is 0.97 for discover the student’s performance. Thus it shows that effective analysis
is performed by using proposed algorithm. From the result, can conclude that proposed system is superior in
performance.
The accuracy is the proportion of true results (both true positives and true negatives) among the total
number of cases examined.
Accuracy can be calculated from formula given as follows
Accuracy =
True positive + True negative
True positive + True negative +False positive + False negative
An accuracy of 100% means that the measured values are exactly the same as the given values.

Figure 7 : Accuracy
From figure 7 can observe that the comparison of existing and proposed system in terms of accuracy
metric. In x axis plot the methods and in y axis plot the accuracy values. In existing scenario, the accuracy
values are lower by using random forest algorithm. The accuracy value of existing scenario is 82 % for discover
the student’s performance. In proposed system, the accuracy value is higher by using the J48 algorithm. The
accuracy value of proposed scenario is 95% for discover the student’s performance. Thus it shows that effective
analysis is performed by using proposed algorithm. From the result, can conclude that proposed system is
superior in performance.
F-Measure is a measure that combines precision and recall is the harmonic mean of precision and
recall, the traditional F-measure or balanced F-score:
Figure 8 : F-Measure
From the figure 8 can observe that the comparison of existing and proposed system in terms of f-
measure metric. In x axis we plot the methods and in y axis plot the f-measure values. In existing scenario, the
f-measure values are lower by using random forest algorithm. The f-measure value of existing scenario is 0.68

for discover the student’s performance. In proposed system, the f-measure value is higher by using the J48
algorithm. The f-measure value of proposed scenario is 0.74 for discover the student’s performance. Thus it
shows that effective analysis is performed by using proposed algorithm. From the result, can conclude that
proposed system is superior in performance.
V. Conclusion
In this proposed system, J48 classification Algorithm is used to classify the student mark based on the
urban and rural. To analyze the academic achievement of urban and rural areas students in order to identify
superior performance is over schooling or in graduation. The analyzing is performed in class wise data. The
most of the students are performing well in their graduation. The B.Sc., Computer Science student data is
trained and B.Sc., Information Technology student data has been tested. It is observed from the experimental
results the Random Forest and J48 Classification Algorithm are shown the higher Precision, recall, accuracy and
f-measure values. The Proposed J48 Classification Algorithm is superior performance for all metrics than the
other algorithm. From the result, the Proposed J48 Classification Algorithm is better for providing efficient
performance.
References
[1] Amershi, S., and Conati, C., (2009) “Combining unsupervised andsupervisedcassificationtobuild user models for exploratory
learning environments” Journal of Educational Data Mining. Vol.1, No.1, pp. 18-71.
[2] Baker, R. S. J. D. "Data mining for education." International encyclopedia of education 7 (2010): 112-118.
[3] Sachin,R.B., & Vijay, M. S, “A Survey and Future Vision of Data Mining in Educational Field”, Paper presented at the
AdvancedComputing & Communication Technologies(ACCT), Second International Conference on 7- 8 Jan. 2012.
[4] Tair, Mohammed M. Abu, and Alaa M. El-Halees. "Mining educational data to improve students’ performance: a case
study." International Journal of Information 2.2 (2012).
[5] Goyal, Monika, and RajanVohra. "Applications of data mining in higher education." International journal of computer science 9.2
(2012): 113.
[6] Abdul Aziz, Azwa, NurHafieza Ismail, and Fadhilah Ahmad."MINING STUDENTS’ACADEMIC PERFORMANCE." Journal of
Theoretical and Applied Information Technology 53.3 (2013): 485-485.
[7] Bhardwaj, Brijesh Kumar, and Saurabh Pal. "Data Mining: A prediction for performance improvement using classification." arXiv
preprint arXiv:1201.3418(2012).
[8] M.Kebritchi,andA.Hirumi,Examiningthe pedagogical foundations of modern educational computer games. Computers and
Education, 5 (4): 1729-1743, 2008
[9] A. McFarlane, N. Roche, and P. Triggs, Mobile Learning:
Research Findings. Becta, July 2007.
http://partners.becta.org.uk/uploaddir/downloads/page_documents/research/mobile_learning_july07.pf (accessed February 4, 2008),
200
[10] MoMath, Mobile Learning for Mathematics: Nokia project in
South Africa. Symbian Tweet, http://www.symbiantweet.com/mobile-learning-

Empirical Study on Classification Algorithm For Evaluation of Students Academic Performance

More Related Content

What's hot (18)

Viewers also liked (11)

Similar to Empirical Study on Classification Algorithm For Evaluation of Students Academic Performance (20)

More from iosrjce (20)

Recently uploaded (20)

Empirical Study on Classification Algorithm For Evaluation of Students Academic Performance