Özet:

Educational Data Mining is an emerging area due to the extensive growth of educational data.  Recently, the amount of data stored in the educational database is increased rapidly. The stored database comprises hidden information about student performance and behavior.  Predicting student’s performance is more challenging due to a huge amount of data in educational databases. Many works carried out their research on data mining techniques to evaluate student performance. But the accurate prediction with minimum consumption is a challenging issue. In order to improve the prediction accuracy, a Regularized Canonical Variate Emphasis Jenks Breaks Boost Data Clustering (RCVEJBBDC) technique is introduced for student grade prediction with lesser time consumption. The RCVEJBBDC technique performs two processes namely Attribute Selection, and clustering for behavior analysis. In the RCVEJBBDC technique, the Regularized Canonical Variate Attribute Selection Process is carried out to select the relevant attributes from the input database using the radial basis kernel function. After attribute selection, Emphasis Jenks Breaks Boost Data Clustering is used to cluster the data based on student behavior analysis. Emphasis Boost Data Clustering combines the weak learner result to form the strong cluster output. Jenks Breaks Cluster (JBC) is considered as the weak learners for clustering the student data. Jenks Breaks is a data clustering technique to group the data values into different clusters.  The Emphasis Boost technique combines the weak learners and provides strong clustering results by minimizing the quadratic error. This in turn helps to improve the student grade prediction accuracy and to minimize the time consumption based on their behavior. Experimental evaluation is carried out for factors such as prediction accuracy, false-positive rate, prediction time, and space complexity with respect to a number of student data. The empirical results demonstrate that the RCVEJBBDC technique provides higher prediction accuracy and lesser prediction time as well as space complexity than the conventional clustering techniques.

Anahtar Kelimeler: