Algoritma Adaboost pada Metode Decision Tree untuk Klasifikasi Kelulusan Mahasiswa

Abstract

Colleges provide higher education as the benchmark of education quality and evaluate higher education syllabi. Graduation rates and enrollment capacity are essential for graduation assessment and decision-making. Unfortunately, some students majoring in statistics failed to finish their studies on time, impacting the accreditation of the study program. It is necessary to examine the characteristics of students who managed and failed to complete their studies on time using the data mining classification method, namely Algorithm C5.0. In this study, Adaboost algorithm and Algorithm C5.0 was employed to classify graduation rates accurately. Graduation data of the Statistics Study Program of Universitas Tanjungpura Batch 1 of 20217/2018 to Batch II of 2022/2023 School years were regarded in this study. First, the entropy, gain, and gain ratio values were measured. After that, each data was given equal weight, and iteration was performed 100 times. The analysis using Algorithm C5.0 showed School Accreditation as the variable with the highest gain ratio, indicating that School Accreditation has the most decisive influence on graduation rates with an accuracy percentage of 70%. This percentage then increased to 82.14% after the boosting using the Adaboost algorithm. Adaboost Algorithm is regarded as good in improving the accuracy of algorithm C5.0. The results of this study can provide insight for colleges in designing policies to increase on-time graduation based on the factors that influence student graduation.

Keywords

• study period
• c5.0 algorithm
• adaptive boosting
• adaboost algorithm