Prediction of Student Graduation with Naive Bayes Algorithm

Abstract

The research carried out in this study is the development and analysis of student performance in the academic field using the Naive Bayes algorithm so that it can help agencies and students see early graduation predictions, and help managers to see the progress and predictions of active student graduation. The purpose of this research is to study student achievement prediction models that have model values. Referring to previous research in reference that getting student prediction results in the form of semester GPA 1,2,3,4, in this study make predictions based on training data and variables that affect the model. The prediction model optimization step by selecting the variable used in the prediction model development is IPS1,2,3,4. The data used in this study are the results of observations from universities. The result of this research is the prediction of Student Achievement Development with Naive Bayes Algorithm based on Ip semester 1.2,3,4 variable and added value are UN rate, Gender, and status stay. From the results of research conducted from the initial stage up to the testing stage the application of the naïve Bayes method for the prediction process of graduate students, it was found that: the application of the naïve Bayes algorithm for model 1 is a model prediction using variable IP students result in an accuracy of 75% and R2 = 68,2%. Model 2 for prediction are used 8 variables namely Nim, Gender, Residence, IPS 1, IPS2, IPS3, IPS4, study period and student status result accuracy of prediction 89% and with a prediction level of R2 = 71,4 %. This certainly improves the performance of the training data efficiency prediction model.