Prathama, Abrarta Fawwas (2025) Pengembangan Sistem Klasifikasi Penyakit Diabetes Menggunakan Xgboost Dan Algoritma Particle Swarm Optimization. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5026211132-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy |
Abstract
Diabetes mellitus adalah salah satu penyakit metabolik yang menjadi penyebab utama kematian dan kesakitan di seluruh dunia, dengan potensi komplikasi serius seperti penyakit
kardiovaskular dan gagal ginjal. Tantangan utama dari penyakit ini terletak pada sifatnya yang kronis dan sering kali tidak terdeteksi pada tahap awal, sehingga diperlukan metode diagnosis dini yang lebih efektif. Penerapan machine learning, terutama algoritma XGBoost yang dioptimalkan dengan Particle Swarm Optimization (PSO), telah terbukti mampu meningkatkan akurasi dalam memprediksi diabetes. Kombinasi antara kemampuan XGBoost dalam klasifikasi
dan efektivitas PSO dalam optimasi parameter model dimanfaatkan untuk mengembangkan sistem prediksi yang dapat mengidentifikasi risiko diabetes dan pra-diabetes secara akurat dan efisien. Tugas akhir ini berhasil mengembangkan sistem prediksi diabetes dan pra-diabetes menggunakan algoritma XGBoost yang dioptimalkan dengan Particle Swarm Optimization (PSO). Sistem ini menunjukkan kinerja luar biasa, dengan model yang mencapai akurasi 100% pada data uji, menggunakan parameter optimal seperti n_estimators 29, max_depth 8, learning_rate 0.050, subsample 0.84, colsample_bytree 0.88, dan gamma 0.43. Penggunaan
PSO secara signifikan meningkatkan akurasi model dan mengoptimalkan hyperparameter, menghasilkan prediksi yang lebih baik. Selain itu, aplikasi web berbasis Flask dikembangkan untuk memungkinkan pengguna memasukkan data medis dan menerima prediksi yang akurat mengenai kondisi diabetes dan pra-diabetes.
====================================================================================================================================================
Diabetes mellitus is a metabolic disease that is a leading cause of death and morbidity worldwide, with potential serious complications such as cardiovascular disease and kidney failure. The main challenge of this disease lies in its chronic nature and the fact that it is often
undetected in the early stages, making effective early diagnosis methods essential. The application of machine learning, especially the XGBoost algorithm optimized with Particle Swarm Optimization (PSO), has proven to improve accuracy in predicting diabetes. The combination of XGBoost's classification capability and PSO's effectiveness in optimizing model parameters is used to develop a prediction system that can accurately and efficiently
identify the risk of diabetes and pre-diabetes. This final project successfully developed a diabetes and pre-diabetes prediction system using the XGBoost algorithm optimized with
Particle Swarm Optimization (PSO). The system demonstrated outstanding performance, with a model achieving 100% accuracy on test data, using optimal parameters such as n_estimators 29, max_depth 8, learning_rate 0.050, subsample 0.84, colsample_bytree 0.88, and gamma 0.43. The use of PSO significantly improved the model's accuracy and optimized hyperparameters, resulting in better predictions. Additionally, a Flask-based web application was developed to
allow users to input medical data and receive accurate predictions regarding their diabetes and pre-diabetes conditions.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Diabetes Mellitus, Machine Learning, Particle Swarm Optimization (PSO), Klasifikasi, XGBoost, Classification, Diabetes Mellitus, Machine Learning, Particle Swarm Optimization (PSO), XGBoost. |
Subjects: | T Technology > T Technology (General) > T174 Technological forecasting |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis |
Depositing User: | Abrarta Fawwas Prathama |
Date Deposited: | 14 Jul 2025 02:04 |
Last Modified: | 14 Jul 2025 02:08 |
URI: | http://repository.its.ac.id/id/eprint/119561 |
Actions (login required)
![]() |
View Item |