@article{article_1635937, title={Predicting Graduation of Associate Degree Students: An Ensemble Learning Approach}, journal={Bilecik Şeyh Edebali Üniversitesi Fen Bilimleri Dergisi}, volume={12}, pages={521–533}, year={2025}, DOI={10.35193/bseufbd.1635937}, author={Kala, Ahmet and Özkurt, Cem and Ok, Yasin}, keywords={On-Time Graduation, Educational Data Mining, Ensemble Learning, Machine Learning}, abstract={On-time graduation is an important indicator of success in associate degree programs that aim to provide students with basic knowledge and skills in a specific profession or field and enable them to quickly enter the workforce. However, graduation delays increase the economic burden and cause psychological difficulties by delaying the process of starting a professional career. This study aims to estimate students’ on-time graduation predictions at an early stage, that is, at the beginning of the semester, and to identify at-risk students who may not graduate on time. Thus, time will be gained for intervention for these students at risk. Ensemble learning methods and classical machine learning models, which were created by combining multiple models, were used for early prediction. The effectiveness of the models was examined with demographic information, high school achievements, university entrance exam results and academic performance data obtained from an associate degree program at a state university. Classification performance estimates were made in three academic stages: the beginning of the first semester, the end of the first semester and the end of the second semester. The results were evaluated according to the F1 score performance metric and it was seen that the LR model performed better at the beginning of the semester and the ensemble learning methods performed better in the other two periods. Additionally, cumulatively adding within-term academic performance data increased the classification performances of all models.}, number={2}, publisher={Bilecik Seyh Edebali University}