Afifi Rufaida, . (2026) DASHBOARD MONITORING DAN PREDIKSI KETERLAMBATAN PEMBAYARAN UKT MENGGUNAKAN ALGORITMA GRADIENT BOOSTING (Studi Kasus: UPN “Veteran” Jakarta). Tugas Akhir thesis, Universitas Pembangunan Nasional Veteran Jakarta.
![[img]](http://repository.upnvj.ac.id/style/images/fileicons/text.png) |
Text
ABSTRAK.pdf Download (666kB) |
|
Text
AWAL.pdf Download (3MB) |
|
Text
BAB 1.pdf Restricted to Repository UPNVJ Only Download (2MB) |
|
Text
BAB 2.pdf Restricted to Repository UPNVJ Only Download (4MB) |
|
Text
BAB 3.pdf Restricted to Repository UPNVJ Only Download (6MB) |
|
Text
BAB 4.pdf Restricted to Repository UPNVJ Only Download (14MB) |
|
Text
BAB 5.pdf Download (1MB) |
|
Text
DAFTAR PUSTAKA.pdf Download (569kB) |
|
Text
RIWAYAT HIDUP.pdf Restricted to Repository staff only Download (169kB) |
|
Text
LAMPIRAN.pdf Restricted to Repository UPNVJ Only Download (2MB) |
|
Text
HASIL PLAGIARISME.pdf Restricted to Repository staff only Download (22MB) |
|
Text
ARTIKEL KI.pdf Restricted to Repository staff only Download (1MB) |
Abstract
Late payment of the Single Tuition Fee (UKT) at UPN “Veteran” Jakarta require structured monitoring so that students at risk of late payment can be identified earlier. This study develops a monitoring dashboard and a late-payment prediction system using the Gradient Boosting algorithm. The dataset comprised 164,355 historical payment records from 2020–2025. After data cleaning, feature engineering, modeling-period selection, train-test splitting, and SMOTE application to the training data, 104,630 records from nine semesters were used for modeling. The best model used an 80:20 split and achieved 77.44% Accuracy, 78.54% Precision, 68.54% Recall, 73.20% F1-score, and 85.04% AUC-ROC. Threshold testing selected 0.28 as the initial operational threshold for a Potentially Late prediction because it was the highest threshold that retained at least 90% Recall, namely 90.29%. The 0.42 threshold was used as the lower boundary of Medium risk because it produced the highest F1-score, while 0.67 was used as the lower boundary of High risk because it was the first threshold to reach at least 90% Precision. The Flask dashboard provides Overview, Monitoring, Analytics, Risk Prediction, and faculty-level risk list features. All 30 Black Box Testing scenarios were valid, and UAT reached 78.33% and was categorized as feasible.
| Item Type: | Thesis (Tugas Akhir) |
|---|---|
| Additional Information: | [No.Panggil: 2310501108] [Pembimbing: Theresia Wati] [Penguji 1: Ika Nurlaili] [Penguji 2: Bobby Suryo Prakoso] |
| Uncontrolled Keywords: | Dashboard, Gradient Boosting, Late Payment Prediction, Monitoring, Single Tuition Fee. |
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software |
| Divisions: | Fakultas Ilmu Komputer > Program Studi Sistem Informasi (DIII) |
| Depositing User: | AFIFI RUFAIDA |
| Date Deposited: | 04 Sep 2026 03:18 |
| Last Modified: | 04 Sep 2026 03:18 |
| URI: | http://repository.upnvj.ac.id/id/eprint/50982 |
Actions (login required)
 |
View Item |