Data Mining in Education: Data Classification and Decision Tree Approach

Abstract

Educational organizations are one of the important parts of our society and playing a vital role for growth and development of any nation. Data Mining is an emerging technique with the help of this one can efficiently learn with historical data and use that knowledge for predicting future behavior of concern areas. Growth of current education system is surely enhanced if data mining has been adopted as a futuristic strategic management tool. The Data Mining tool is able to facilitate better resource utilization in terms of student performance, course development and finally the development of nation's education related standards. In this paper a student data from a community college database has been taken and various classification approaches have been performed and a comparative analysis has been done. In this research work Support Vector Machines (SVM) are established as a best classifier with maximum accuracy and minimum root mean square error (RMSE). The study also includes a comparative analysis of all Support Vector Machine Kernel types and in this the Radial Basis Kernel is identified as a best choice for Support Vector Machine. A Decision tree approach is proposed which may be taken as an important basis of selection of student during any course program. The paper is aimed to develop a faith on Data Mining techniques so that present education and business system may adopt this as a strategic management tool. decision tree approach and decision rule approach. Considering the global opportunities coupled with global competition even in case of education it is essential to admit the best students as far as possible. So their academic performance and subsequent placements are best in the world. Data mining is useful whenever a system is dealing with large data sets. In any education system, student records i.e. enrollment details, course eligibility criteria, course interest and academic performance may be an important consideration to analyze various trends since all the systems are now computer based information system so data availability, modification and updation are a common process now. Data warehousing may be taken as good choice for maintaining the records of past history. The data warehouse can be easily developed in any education institute with the adaptation of common data standard. Common data standards may eliminate the need of data clarity and modification before loading this for a data warehouse. An institute with efficient Data Warehousing and Data Mining approach can find out novel way of improving student's behavior, success rate and course popularity. All these effort may finally improve the quality of education, better student intake, better career counseling and overall practices of education system. In Data Mining classification clustering and regression are the three key approaches. Classification is a supervised learning approach in which students are grouped into identified classes (1). Classification rules may be identified from a part of data known as training data and further it may be tested for rest of the data (2). The effectiveness of classification approach may be evaluated in terms of reliability of the rule with test data set. Clustering approach is based on unsupervised learning because there are no predefined classes. In this approach data may be grouped together as cluster (2), (3). The usability of clusters in terms of relevant area may be interpreted by data mining expert. Regression is a data mining approach in which it uses the explanatory variable to predict an outcome variable. For example, performance appraisal of faculty members may be done by regression analysis. Here, faculty qualification, feedback rating, amount of content covered may be taken as explanatory variable and faculty salary, increment, bonus and perks may be estimated as outcome variable so regression may be the best way to setting few important parameters based on existing variables (2).