Comparative Analysis of Random Forest and Gradient Boosting for Early Detection of Dengue Fever in Regional Hospital

Wiryani, Made Dwi Aprillia Kusuma (2026) Comparative Analysis of Random Forest and Gradient Boosting for Early Detection of Dengue Fever in Regional Hospital. Undergraduate thesis, Universitas Pendidikan Ganesha.

[img] Text
2215091106-COVER.pdf
Download (432kB)
[img] Text
2215091106-ABSTRAK.pdf
Download (206kB)
[img] Text
2215091106-BAB 1 PENDAHULUAN.pdf
Download (322kB)
[img] Text
2215091106-BAB 2 KAJIAN TEORI.pdf
Restricted to Repository staff only
Download (590kB) | Request a copy
[img] Text
2215091106-BAB 3 METODOLOGI PENELITIAN.pdf
Restricted to Repository staff only
Download (504kB) | Request a copy
[img] Text
2215091106-BAB 4 HASIL DAN PEMBAHASAN.pdf
Restricted to Repository staff only
Download (978kB) | Request a copy
[img] Text
2215091106-BAB 5 PENUTUP.pdf
Restricted to Repository staff only
Download (183kB) | Request a copy
[img] Text
2215091106-DAFTAR PUSTAKA.pdf
Download (211kB)
[img] Text
2215091106-LAMPIRAN.pdf
Download (1MB)

Abstract

Dengue Fever (DF) is a disease caused by the dengue virus (DENV), primarily transmitted through the bites of infected mosquitoes. Dengue surged globally in 2024 with 7.6 million dengue cases reported by the World Health Organization (WHO). Dengue fever shares overlapping symptoms with other febrile conditions, such as febris, influenza, chikungunya, etc., commonly presenting with high body temperature, making early detection difficult without prior laboratory examination. Machine learning offers a way to support early detection through classification. This study compares two algorithms, Random Forest and Gradient Boosting, both well-known for classification tasks, evaluated using accuracy, precision, recall, and F1-score. Recall is prioritized as the primary metric to minimize false negatives in disease classification. The dataset was obtained from RSUD Buleleng through an ethical procedure, consisting of 212 entries, with 211 cleaned records and 11 selected clinical features used in modelling. Stratified 5-fold cross-validation and GridSearchCV applied to ensure both classes preserved in training and testing. Under stratified 5-fold cross-validation, RF maintained its advantage with a mean CV recall of 87.32%, slightly higher than GB's mean CV recall of 81.76%, leading to RF being selected for final model deployment, implemented via a React.js and Flask API.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Classification, Dengue, Gradient Boosting, Random Forest, Recall
Subjects: T Technology > T Technology (General)
Divisions: Fakultas Teknik dan Kejuruan > Jurusan Teknik Informatika > Program Studi Sistem Informasi (S1)
Depositing User: Made Dwi Aprillia Kusuma Wiryani
Date Deposited: 15 Jun 2026 08:47
Last Modified: 15 Jun 2026 08:47
URI: http://repo.undiksha.ac.id/id/eprint/29754

Actions (login required)

View Item View Item