EDUCATIONAL DATA MINING FOR STUDENT ACADEMIC PREDICTION USING K-MEANS CLUSTERING AND NAÏVE BAYES CLASSIFIER

Abstract

This study proposes the merging of the K-Means clustering data mining method and the Naïve Bayes classifier (K-Means Bayes) for better results in data processing for Student Academic Performance data. Data was taken from the Student Academic Performance dataset which is used as a test case. The amount of data used in this study were 131 data and 21 attributes. The accuracy of the results obtained from the combination of the proposed method is 97.44%. The results obtained when compared with calculations using the K-Means method and calculations using the Naïve Bayes method, the proposed method (K-Means Bayes) gives better results. Although the initial centroid determination on the K-Means method is done randomly, the impact can be reduced by adding the Naive Bayes classifier method which results in a better accuracy value, thereby increasing the accuracy of the method used. Compared to the K-Means and Naïve Bayes methods, the proposed method increases the accuracy of about 27% of the Naïve Bayes algorithm and about 23% of the K-Means algorithm. With the results obtained, it can be concluded that the proposed method can improve predictions of student academic performance data. The initial centroid determination for grouping in the K-Means method can affect the quality of the accuracy of the data produced