Analisis Prediksi Kelulusan Mahasiswa Menggunakan Decission Tree Berbasis Particle Swarm Optimization

Abstract

Good accreditation results are the goal of the college. With good accreditation, prospective students can glance at and enter the tertiary institution. To achieve this, there are several aspects that affect good accreditation results, one of which is graduate students who play an important role in determining accreditation. Timely graduate students can benefit the college or a student. Graduates can be predicted before the final semester using a method one of which is the decision tree. Decision tree is a method that is simple and easy to understand by producing rules in the form of a decision tree, but using a decision tree model alone is not enough to produce optimal results. So we need a method for optimization that is particle swarm optimization with advantages can improve accuracy by eliminating unused features. From the results of research with primary data of 2000-2003 graduate students in Amik PPMI Tangerang explained that the particle swarm optimization method can increase accuracy by 87.56% and increase by 01.01% from the decision tree method with a value of 86.55%. From the particle swarm optimization method can also find out which unused attributes have no weight, so that way can improve accuracy. From the results of the increase, it can be used by the Amik University of Tangerang to prevent students from graduating on time.