Abstract-Educationaldata mining techniques are widely used in the educational system. They help theinstitute in improving the performance of student and retrieving informationfor many useful purposes. This review discusses a journal paper on the Evolutionaryalgorithm for generating personalized system in educational data mining.
We discoursethe methodology and implementation of the system in the study in detail. Furtherwe present the critical review and limitation of the study. we also deliberate thesuitable solution of the critics in the end. I.
INTRODUCTIONEducational data mining(EDM)is a vast research field related to machine learning, data mining and statisticalsciences. EDM help to retrieve information from tons of data in advance andintelligent educational systems. The EDM techniques extract meaningfulinformation from large databases based on the need of tutor.
They help in improvementof student’s activities and learning ability. The raw data can be all librarybooks or students’ past results. The EDM techniques assist the student, tutoror administration user to get the suitable material and process it for furtheruse. Many systems like predicating student’s performance, failure and finalgrade, automated quiz systems and generating students group use EDM techniquesto improve their performance and achieve their goals. Evolutionary algorithm isa genetic algorithm in artificial intelligence which work well in predication andevolution of results. Many researches have been using evolutionary algorithm indata mining to achieve maximum productivity. Evolutionary algorithms areactually used in biological prediction systems but a number of researchers are implementingin educational data mining. Many researches concluded they work better from greedyapproach in data mining.
The review presents a briefreview on the use of evolutionary algorithm in educational data mining. Differenttypes of evolutionary algorithm are genetic programming, differential algorithmand genetic algorithm. The review answers the following research questions:1.
Does the research article define what It claimsin the abstract?2. Are the techniques and methodology used efficientenough?3. Are the results properly defined and concluded? II. EVOLUTIONARY ALGORITHMINEDUCATIONALDATAMININGThe Study presents arebrief detail of the educational data mining and where they are used in realworld.
The data mining techniques are applied to retrieve information fromlarge databases of education institutes and process for the improvement oftutor and student performance. The real-world applications of educational datamining are predicting student performance and grades through the past results ofthe student. Improvement of domain and student models, research into studentsand learning material, predicting in the learning tools are some areas ineducational environment. These tools can be used tutors, students, courseauthor, institute and administration. Further the study defines the as that canbe assigned to educational systems that use educational data mining. Some tasksare:1) Getting feedback for tutor.2) Recommending suitable course to thestudents.3) Making students’ group.
4) Predicting the performance of studentClustering, classification,and regression techniques are implemented in these system for data mining. Themajor task of these systems is help the instructor and student to develop theircourseware. The Evolutionary algorithm used in EDM have helped to improve itsworking. The study shows that it has proved to be better than otherclassification techniques. The study further presents a summary of the survey.The educational systems depend on four parts: 1) the main task2) users3) EDM technique4) the algorithm.
III. METHODOLOGYThe goal of the study wasto present an efficient evolutionary algorithm which is a solution to EDM problems.It categorizes the EDM related journal papers in a table. It consists of thecourseware construction and maintenance problem as the main goal in the study. Theresearch also shows that the evolutionary algorithm has tend to work betterthan other data mining techniques in EDM. They choose course authors as theirmain focus in the study as the recommendation for them is an important aspect.
Table I shows 12 journalpapers that used evolutionary algorithm for their system implementation andinformation extraction. The table is divided into four columns as the educationalsystem parts defined in section 2. Furthermore, the fig 1 is used to sketch thenumber of publication related to the tasks mentioned in the table. The graph showsthat most of the journal paper used evolutionary algorithm for getting the feedbackfor the tutor. On the other hand, from table we can easily conclude thatmajority papers used genetic algorithm for their task.The study than deliberatethe techniques of courseware construction and a literature survey on thisissue. It introduces many systems such as forming web tutor tree and cluster ofstudents for predicting their performances through the educational history. Also,many articles for recommendation of the course to students are reviewed in thestudy.
many articles related to improvement if learning style are also reviewedthat used document index graph, association rule mining and other learningstyles. The author presents the benefits of using these techniques which canhelp in the improvement of the EDM systems. The study used KEELsoftware for the generation of their results. The implementation of the geneticalgorithm is used to find the students performance. It shows the average ofsupport value is 0.36 whereas the confidence value is 1. Which concludes thatif the students lie between the given value than he is a sensitive learner. Thesample courseware construction can be implemented to get better results for students’performance prediction.
IV. CRITICAL EVOLUTIONThe study presents abrief discussion of the evolutionary algorithm in educational data mining. The real-worldapplications, different types of users and their tasks, three types ofevolutionary algorithm and the four parts of educational systems and theirproblems are defined in detail in the study.
It also presents a briefliterature survey on the key area. The study also categorized the journalpapers in table and show the number of publications in a graph.The table shows the journalarticles on only genetic algorithm and programming while it claims to have a surveyon Evolutionary algorithm and no other type of EA. It means that the study isnot conducted overall on the evolutionary algorithms. While many previousresearch articles show that differential algorithms prove to be the best in theEDM systems. Further it can be clearly seen from the table that no journal hasbeen selected from the clustering techniques. Also, the study reviews only twelvejournal papers for this objective.
The paper also notdiscusses the table and figure in detail and what they conclude from them. Thestudy doesn’t deliberate the main goal line of using these presentationmethods. The study states all the evolutionaryalgorithm in detail and defines all its factors.
We can see that no limitationsand drawbacks of the algorithms are defined. The journal papers are only viewedand the focus was only on their data mining techniques. while their drawbacksand solution can be discussed in the paper.
The definitions are not enough fora survey paper.The paper also defines noselection criteria and research questions for the journal article. The paperjust defines the techniques used in the systems of different journal papers.
The author claims theobjective of their study to find an efficient evolutionary algorithm for theEDM problem in section 4. While no where else they mention or conclude thisobjective and not discuss the best algorithm in this perspective.