IMPLEMENTASI DIRICHLET SMOOTHING PADA NAÏVE BAYES UNTUK KLASIFIKASI KINERJA AKADEMIK MAHASISWA UNIVERSITAS MADURA

Abstract

This study focused on evaluating the implementation of Dirichlet Smoothing on Naïve Bayes in predicting students' graduation at Madura University. Evaluation on the performance of students conducted in the second and fourth year so that the academics can know and take action in the form of warning or warning to students to improve learning outcomes so as to pass on time. The measurement indicators used are GPA, credits that have not been taken and the length of study. Evaluations performed in the second year using established standard values, while classification with the Naïve Bayes algorithm, are used in the evaluation of the speedy year. To improve the accuracy of classification results in Naïve Bayes, Dirichlet Smoothing is used. By comparing the results of Naïve Bayes classification performance with Dirichlet Smoothing and without Smoothing, it was found that the results of classification calculation for 5 times in 725 training data and 200 testing data taken at random, then obtained the difference of accuracy value reached 5.48%, the precision value of 7.1% and the recall value is 4%. In other words, the addition of smoothing methods can improve the performance of Naive Bayes.