Memprediksi Kelulusan Mahasiswa: Graduate dan Dropout dengan Support Vector Machine dan GridSearchCV

Abstract

In today's educational landscape, having a model to predict whether a student will graduate or drop out based on their academic statistics is highly beneficial. Such a model allows for early assessment of academic success. Human calculations alone can be time-consuming and often lack accuracy, hence the introduction of machine learning models to address this issue. This research utilizes a dataset comprising undergraduate students from various majors in higher education institutions. The data were collected while the students were still enrolled, with their grades from the first year serving as a key feature. The response variable in the dataset is labeled as either 'dropout' or 'graduate'. We employ Support Vector Machines (SVM) with GridSearchCV optimization to build the predictive model. The goal of this model is to predict a student’s academic success as early as their first-year statistics are available. If a student is predicted to drop out, targeted interventions can be provided to help them overcome challenges, ultimately aiming to improve graduation rates.

Keywords: siswa, akademik, dropout, graduate, SVM, hyperparameter tuning, klasifikasi, prediksi, machine learning, GridSearchCV