<![CDATA[College is a place for students to pursue higher education. Both state and private universities compete to be the best universities to produce the best graduates. The number of student graduates is an indicator of the success of a higher education institution, which will have an impact on government accreditation and public assessment. The aim of this research is to predict student graduation in order to know whether they will graduate on time or late by applying data mining techniques, namely classification using the C4.5 algorithm to obtain patterns of student graduation delays and the particle swarm optimization (PSO) algorithm to increase the accuracy of the C4 algorithm. 5. Testing uses cross validation tests, confusion matrix and ROC curve. The results of this research are that the C4.5 algorithm with particle swarm optimization (PSO) has an accuracy value of 86.72%, which is better than the C4.5 algorithm, whose accuracy is 82.05% and the difference between them is 4.67%. The difference between the AUC value of 0.033 was obtained from the C4.5 algorithm model, which had an AUC value of 0.870 with a good classification diagnostic level, and the C4.5 algorithm with PSO had an AUC value of 0.903 with an excellent classification diagnostic level. IPS3 is the attribute that most influences the accuracy of student graduation. The results of the C4.5 algorithm rule with PSO can be applied to create applications for GUI-based student graduation predictions.]]>
Copyrights © 2023
