Comparison Of Data Mining Methods For Recipient Prediction Poor Student Assistance (BSM) In MAN 2 North Lampung

Abstract

MAN 2 North Lampung is a State Madrasah Aliyah or equivalent to Senior High School which has implemented poor student assistance (BSM),  system by considering the economic condition of the students or the criteria that have been set. Selection of  BSM acceptance is a semi-structured problem type meaning that this process is not a routine agenda of a school but the agenda held at a certain time that is when students are in class X. Determined the BSM recipient candidate must collect the data file of candidate selection of BSM recipients from students’ data coming from poor family to very poor family. So it takes a relatively long time, as well as high accuracy in making decisions. In predicting students who receive BSM, the authors apply the data mining process using the Naive Bayes method, Decision Tree, K-NN. The attributes used consist of siblings, Parent Occupation, parental income, smart indonesian card (KIP) recipients or not, the status of the family as an orphan or not. To perform the process of data mining in need of tools aids that is RapidMiner 5. The Implementation of data mining using a comparison of  3 methods can be seen based on the sample number of 393 students. the results of the precision value of the Naive Bayes method are better used for this study compared to other methods. While based on recall and accuracy values, Decision Tree is better used than other methods. But when viewed from the overall results of BSM receiver predictions, the most influential variable is parent income and receiver the KIP card.

Keywords: Data Mining , Decision Tree, Naive Bayes, and K-NN