Sandi, Putra Ari and Arvio, Yozika and Karmila, Sely (2025) PERBANDINGAN AKURASI MODEL XGBOOST, SVM, DAN LOGISTIC REGRESSION DALAM MEMPREDIKSI DATA PENYAKIT DIABETES. Diploma thesis, ITPLN.
202131016_Putra Ari Sandi_Revisi_Skripsi_Putra Ari Sandi.pdf
Restricted to Registered users only
Download (9MB)
Abstract
Diabetes merupakan penyakit kronis yang membutuhkan deteksi dini untuk mencegah komplikasi serius. Perkembangan teknologi kecerdasan buatan, khususnya algoritma machine learning, memberikan peluang dalam membangun sistem prediksi penyakit yang lebih akurat. Penelitian ini bertujuan untuk membandingkan performa tiga algoritma klasifikasi, yaitu Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), dan Logistic Regression (LR), dalam mengidentifikasi kemungkinan seseorang menderita diabetes. Dataset yang digunakan berasal dari Puskesmas Kembangan (2025), berisi rekam medis pasien tahun 2024 dengan lima fitur utama: Usia, Jenis Kelamin, Tekanan Darah, Indeks Massa Tubuh (IMT), dan Kadar Glukosa. Evaluasi dilakukan menggunakan metrik akurasi, presisi, recall, dan F1-score. . Hasil menunjukkan bahwa XGBoost memiliki performa terbaik dengan akurasi 99.84% dan, yang paling berpengaruh, nilai recall yang mendekati 100%. Kemampuan recall ini sangat krusial untuk deteksi dini diabetes karena mengukur seberapa baik model dapat mengidentifikasi semua pasien yang benar-benar menderita diabetes. Nilai recall yang tinggi pada XGBoost secara efektif meminimalkan False Negative, yaitu kesalahan yang paling berbahaya dalam diagnosis medis karena pasien tidak terdeteksi dan tidak mendapatkan penanganan. Kinerja XGBoost ini jauh melampaui SVM akurasi 89.67% dan LR akurasi 89.61%. Perbedaan performa mencerminkan karakteristik unik masing masing model. Temuan ini diharapkan dapat mendukung pengembangan sistem pendukung keputusan medis yang akurat dan efisien untuk deteksi dini diabetes di layanan kesehatan.
Diabetes is a chronic disease that requires early detection to prevent serious complications. The development of artificial intelligence technology, especially machine learning algorithms, provides opportunities to build more accurate disease prediction systems. This study aims to compare the performance of three classification algorithms, namely Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), and Logistic Regression (LR), in identifying the likelihood of a person suffering from diabetes. The dataset used comes from the Kembangan Health Center (2025), containing the patient's medical records in 2024 with five main features: Age, Gender, Blood Pressure, Body Mass Index (BMI), and Glucose Levels. Evaluations were conducted using accuracy, precision, recall, and F1-score metrics. The results show that XGBoost has the best performance with an accuracy of 99.84% and, most influentially, a recall value that is close to 100%. This recall capability is crucial for early detection of diabetes because it measures how well the model can identify all patients who actually have diabetes. The high recall value of XGBoost effectively minimizes False Negatives, which are the most dangerous errors in medical diagnosis because patients are not detected and do not receive treatment. The performance of this XGBoost far exceeds the SVM accuracy of 89.67% and the accuracy LR of 89.61%. The difference in performance reflects the unique characteristics of each model. These findings are expected to support the development of accurate and efficient medical decision support systems for early detection of diabetes in healthcare.
| Item Type: | Thesis (Diploma) |
|---|---|
| Uncontrolled Keywords: | Diabetes, Extreme Gradient Boosting, Logistic Regression, Prediksi Kesehatan, dan Support Vector Machine Diabetes, Extreme Gradient Boosting, Logistic Regression, Health Prediction, and Support Vector Machine |
| Subjects: | Skripsi Bidang Keilmuan > Teknik Informatika |
| Divisions: | Fakultas Telematika Energi > S1 Teknik Informatika |
| Depositing User: | Sudarman |
| Date Deposited: | 13 Oct 2025 08:21 |
| Last Modified: | 13 Oct 2025 08:21 |
| URI: | https://repository.itpln.ac.id/id/eprint/2166 |
