The Study of Neural Network Algorithm, Random Forest for Classification of Student Graduation

Abstract

The academic performance is one of aspect which has remained the bechmark of the success in learning activities at an university. The indicator of academic performance in the university is the students able to complete their studies on time. Unfortunately, the problem regarding academic performance was associated with the completion time of student studies in Faculty of Economics, University of Garut. In this research explore the model that able to classify the graduation of student through the data mining classification technique by comparing the Neural Network Algorithm dan Random Forest. The classification conducted by evaluating the academic performance based on Semester Performance Index (IPS) first years in the beginning and use the demographics of students as attributes that will be used in the dataset. Based on the results of several model tests from the data train, totaling 1467 data records and 25 attributes. It shows that the 14th Random Forest test model produces a Recall Performance value of 72.70%, 74.70% for Accuracy Performance, 72.80% for Precision Performance and 74.70% for F-measure Performance.