EARLY WARNING SYSTEM KELULUSAN MAHASISWA MENGGUNAKAN ALGORITMA RANDOM FOREST

Tedut, Elcky Mardiantho (2025) EARLY WARNING SYSTEM KELULUSAN MAHASISWA MENGGUNAKAN ALGORITMA RANDOM FOREST. S1 Teknik Informatika thesis, STMIK Widya Cipta Dharma.

	Text 2143045-S1-Teknik Informatika_compressed (1).pdf Restricted to Repository staff only Download (1MB) | Request a copy
	Text 2143045-S1-Jurnal.pdf Download (1MB)

Abstract

Kelulusan tepat waktu merupakan indikator penting dalam menilai keberhasilan suatu institusi pendidikan tinggi. Di STMIK Widya Cipta Dharma, rata-rata masa studi mahasiswa mencapai 4,7 tahun, menunjukkan bahwa masih banyak mahasiswa yang belum lulus tepat waktu. Fenomena ini berpotensi memengaruhi reputasi institusi dan akreditasi program studi. Penelitian ini bertujuan untuk membangun Early Warning System (EWS) menggunakan algoritma Random Forest guna memprediksi status kelulusan mahasiswa, apakah Tepat Waktu, Terlambat, atau Drop Out (DO). Data yang digunakan meliputi IPK ,IP Semester 1 hingga 6, jumlah SKS yang telah ditempuh, semester,bobot, dan Status Pembayaran. Metode pengumpulan data dilakukan melalui studi pustaka, wawancara, dan dokumentasi untuk memperoleh data akademik. Hasil penelitian menunjukkan bahwa sistem yang dibangun mampu mengklasifikasikan mahasiswa dengan tingkat akurasi sebesar 99%. Selain itu, dilakukan implementasi rule extraction dari model Random Forest untuk memperoleh aturan-aturan klasifikasi secara eksplisit. Aturan ini memberikan interpretasi logis terhadap keputusan model dan menjadi dasar dalam pengambilan kebijakan akademik. Website yang dikembangkan memungkinkan proses klasifikasi dilakukan secara manual maupun massal melalui file Excel, serta menampilkan hasil klasifikasi secara interaktif dan informatif. Dengan adanya sistem ini, pihak kampus dapat melakukan tindakan preventif lebih awal terhadap mahasiswa yang berisiko tidak lulus tepat waktu. Namun perlu diigat bahwa akurasi tinggi bukanlah segalanya,ini justru mengindikasikan data Leakage. ============================================================ On-time graduation is an important indicator in assessing the success of a higher education institution. At STMIK Widya Cipta Dharma, the average study duration for students is 4.7 years, indicating that many students are still not graduating on time. This phenomenon has the potential to affect the institution's reputation and the accreditation of its study programs. This research aims to develop an Early Warning System (EWS) using the Random Forest algorithm to predict students’ graduation status whether On Time, Late, or Drop Out (DO). The data used includes GPA, Semester GPA from the first to the sixth semester, the total number of completed credits, semester, weight, and payment status. Data collection methods involved literature study, interviews, and documentation to obtain academic data. The results show that the system developed is capable of classifying students with an accuracy rate of 99%. Additionally, rule extraction was implemented from the Random Forest model to explicitly derive classification rules. These rules provide logical interpretations of the model’s decisions and serve as a basis for academic policy-making. The developed website allows for both manual and bulk classification using Excel files, and it displays classification results in an interactive and informative manner. With this system, the campus can take early preventive action for students at risk of not graduating on time. However, it is important to note that high accuracy is not everythingthis actually indicates data leakage.

Item Type:	Thesis (S1 Teknik Informatika)
Additional Information:	Pembimbing 1 : Wahyuni, S.Kom., M.Kom. Pembimbing 2 : Pitrasacha Adytia, S.T., M.T.
Uncontrolled Keywords:	Early Warning System, Random Forest, Prediksi Kelulusan, Klasifikasi, Rule Implementation, STMIK Widya Cipta Dharma
Subjects:	Q Science > QA Mathematics > QA76 Computer software
Divisions:	Teknik Informatika
Depositing User:	Mr Elcky Mardiantho Tedut
Date Deposited:	08 Aug 2025 05:31
Last Modified:	08 Aug 2025 05:31
URI:	http://repository.wicida.ac.id/id/eprint/6237

Actions (login required)

View Item