Enhanced E-Learning Using Data Mining A Rule Based Classification Approach
Enhanced E-Learning Using Data Mining A Rule Based Classification Approach
Enhanced E-Learning Using Data Mining A Rule Based Classification Approach
In-depth analysis
Index Terms e-learning, quiz, data mining, faculty, In-depth analysis can be done on results to extract useful
student, PACR, recommendations, classification, rules, information about students.
performance level, grades, analysis, NNGE, WEKA, score.
Eco-friendly
Going from hard-copy tests/ quizzes to offering the same
I. INTRODUCTION capabilities online reduces consumption of goods such as
Due to a tremendous advancement in Internet technology paper.
and World Wide Web, the concept of online learning has
become quite popular among students and teachers Powerful self-assessment tool
throughout since last decade. As a result, online learning Testing and quizzing online will usually provide the user with
systems are extensively used in schools and colleges results instantly. This is good for students because it allows
integrated with traditional classroom coaching to promote them to know what they did wrong immediately, what they
learning and improved understanding of concepts using need to focus on, and how to improve should they have to
audio/visual aids. These online learning systems consist of retake the test.
audio & video lessons, tutorials, online books, text lessons,
journals, short quizzes & exams. All these contents are easily However the quizzes available online are inadequate as
accessible to students on demand. they fail to provide a comprehensive analysis of results. Such
Online quizzes and exams are one such popular and vital inadequate quizzes only notify to student, the score achieved
form of online learning tools. It provides an array of benefits and the questions which went right/wrong.
to both students as well as teachers. Some of the advantages of
online quizzes are as follows: The task of the proposed system is to take existing system
to the next level. In proposed system, the quiz section is
Reduced efforts divided into different subjects and each subject is further
E-learning relieves the need for testing to be done at a specific divided into 4 Modules wherein each module consists of quiz
hour, but it also makes testing a hassle-free task as corrections questions related to topic. Every question in quiz has a text
are automated. In the cases of "Essay Question" tests, explanation, video explanation, reference link of the most
appropriate study material on web and PDF content
Manuscript received April 24, 2015. associated with it. The scores of a student for all the modules
Gaurav Kothari, Department of Computer Engineering, KJ Somaiya of a respective subject will be analysed and fed into
College of Engineering, Vidyavihar, Mumbai, India
Anup Ahuje, Department of Computer Engineering, KJ Somaiya College
classification algorithm which will predict performance level
of Engineering, Vidyavihar, Mumbai, India of the student. If the performance of the student is poor, then
Amish Patel, Department of Computer Engineering, KJ Somaiya College the respective faculty and student will be notified and
of Engineering, Vidyavihar, Mumbai, India recommendations for study material will be provided to
Abhinandan Khilari, Department of Computer Engineering, KJ Somaiya
student by system as well as faculty. Further the proposed
College of Engineering, Vidyavihar, Mumbai, India
Prof. Jyothi Rao, Department of Computer Engineering, KJ Somaiya system can be used in schools and colleges along with
College of Engineering, Vidyavihar, Mumbai, India classroom coaching to enhance learning experience.
501 www.erpublication.org
Enhanced E-Learning Using Data Mining a Rule Based Classification Approach
502 www.erpublication.org
International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869, Volume-3, Issue-4, April 2015
Registered Students can perform following tasks Quiz section is divided into Subjects and each subject
consists of 4 modules. Each module has a pool of easy,
1. Access the Online quiz by selecting respective subject medium and hard questions. Student has to select subject and
and module. then chapter to start quiz consisting of 15 questions
2. Student can check their performance containing easy, medium and hard questions (5 each) picked
module-wise/subject-wise and overall rank among in random order. After submitting the quiz, a detailed report
other students easily with help of graph and pie will be provided to student consisting of following
charts. 1. Grade ( A , B , C )
3. If student is performing poor in particular 2. Questions Correct & Wrong
chapter/subject, he/she will be automatically 3. Text Explanation for every question
provided recommendations for learning material 4. Video explanation for every question
from system as well as respective subjects faculty. 5. Reference link of online learning material
4. Students can email their queries directly to respective 6. PDF/Image files
subjects faculty. Responses will be provided within
24 hours. Student has to specify the subject name Quiz Scoring
and topic followed by his doubt. Each module consists of 15 multiple choice quiz questions
5. Student can provide his valuable feedback about the containing 5 easy, 5 medium and 5 hard questions in random
application. order. Easy question has 1 point, medium question has 2
points and hard questions have 3 points each. Hence Total
B. Faculty Module score is 30.
Faculty module consists of faculty entity. The proposed Difficulty Points Questions Total Points
system is faculty controlled application hence faculty has vital Easy 1 5 1 x 5 =5
role in working of application. Faculty has to register to work Medium 2 5 2 x 5 =10
on application. Following attributes of faculty are taken into Hard 3 5 3 x 5 =15
consideration during registration.
Total Points for a quiz = 5 +10+ 15 = 30
Attribute Name Description Table 3. Total Score of each quiz
Faculty_ID Id of Faculty Now if a student get x easy, y medium and z hard questions
Faculty_Name Name correct out of 15 questions, then score in module i is of subject
Faculty_Subject Subject which faculty Teaches is calculated as
Email Institutional Email address
Contact Contact Number
Table 2. Faculty Registration
Where Si is students performance in module i for a particular
subject. Subsequently, Si values for all modules in subject are
Faculty performs following Task calculated and converted into grades using below table.
1. Faculty can add/modify/delete quiz questions.
Grades Value Si Value out of 30
2. Faculty can monitor performance of students
A Above 25
module-wise/subject-wise easily with help of graph
B 15-24
and pie charts.
C Below 15
3. If student is performing poor in particular
chapter/subject, respective subjects faculty will be Table 4. Grades Classification
alerted so that faculty can provide necessary tips and
recommend best study content through email. Attribute Name Description
4. All of the student queries/doubts can be answered by Question_ID Specifies ID of Question
faculty through email. Course_Name Name of Subject
5. Faculty can assess the feedback of students. Chapter_Name Name of Chapter
Question_Text Question
C. Quiz Module Question_Type Specifies question Type
Easy/Medium/Hard
Quiz module consist of multi-course quiz. Quiz section is First First Option
divided into subjects and each subject consists of multiple Second Second Option
modules. Each module has pool of questions. The Questions Third Third Option
of the quiz are added into the database by faculty only. Fourth Fourth Option
Final Correct Option
Creation of Quiz First/Second/Third/Fourth
Quiz questions are created, deleted or modified by faculty Reference_Link Link of the study material
members using faculty panel. File_Path File path of the Study
material Video /Image/PDF
Working of Quiz
Table 5. Quiz database attribute
503 www.erpublication.org
Enhanced E-Learning Using Data Mining a Rule Based Classification Approach
Nearest-Neighbor-Like Algorithm
Nearest neighbour classifiers are a class
of non-parametric methods used in statistical classification
(or pattern recognition)[6]. The method classifies objects
based on closest training examples in the feature space.
Among the various methods of supervised statistical pattern
recognition, the Nearest Neighbour rule achieves consistently
high performance, without a priori assumptions about the
Figure 2. Development phases distributions from which the training examples are drawn. It
involves a training set of both positive and negative cases. A
A. Requirement gathering and analysis new sample is classified by calculating the distance to the
nearest training case; the sign of that point then determines the
Requirement gathering and analysis is critical in determining classification of the sample. The k-NN classifier extends this
the success of this application. Before the system idea by taking the k nearest points and assigning the sign of
development, problems and data understanding is performed the majority. It is common to select k small and odd to break
to define the project goal and objectives. The problems of the ties (typically 1, 3 or 5). Larger k values help reduce the
existing systems are identified and analysed for its effects of noisy points within the training data set, and the
effectiveness and efficiency in term of functionality. After the choice of k is often performed through cross-validation.
problems are identified, the solutions to solve each problem is
identified and collected through literature survey. Besides,
other similar systems are studied and analysed for its features,
strengths and weaknesses. This helps to identify the needs and
opportunities for the proposed system.
504 www.erpublication.org
International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869, Volume-3, Issue-4, April 2015
VI. CONCLUSION
V. FUTURE WORK
Future work includes addition of online tutorials and
assignment in addition to quiz so that more advanced
classifiers like ID3 can be applied to get more accurate
prediction results. Further the system can be used on
commercial scale and the respective response and operational
efficiency can be analysed.
505 www.erpublication.org