IPB University Logo

SCIENTIFIC REPOSITORY

IPB University Scientific Repository collects, disseminates, and provides persistent and reliable access to the research and scholarship of faculty, staff, and students at IPB University

AI Repository
 
Building and Categories


      View Item 
      •   IPB Repository
      • Final Assignments
      • Undergraduate Final Assignments
      • UF - School of Data Science, Mathematic and Informatics
      • UF - Statistics and Data Sciences
      • View Item
      •   IPB Repository
      • Final Assignments
      • Undergraduate Final Assignments
      • UF - School of Data Science, Mathematic and Informatics
      • UF - Statistics and Data Sciences
      • View Item
      JavaScript is disabled for your browser. Some features of this site may not work without it.

      Peningkatan Kinerja XGBoost dalam Klasifikasi Peubah Penting Indeks Prestasi Mahasiswa SNBP IPB University Berbasis EasyEnsemble

      Thumbnail
      View/Open
      Cover (497.3Kb)
      Fulltext (879.7Kb)
      Lampiran (219.3Kb)
      Date
      2026
      Jenis/Type
      Skripsi
      Subtype
      Undergraduate Theses
      Author
      FAYIZA, SALSABILA
      Syafitri, Utami Dyah
      Firdawanti, Aulia Rizki
      Metadata
      Show full item record
      Abstract
      Implementasi Kurikulum Merdeka menekankan pembelajaran yang fleksibel dan berbasis kompetensi, sehingga menghasilkan mahasiswa dengan latar belakang akademik yang semakin beragam. Kondisi ini berkaitan erat dengan skema penerimaan SNBP yang menyeleksi mahasiswa berdasarkan prestasi akademik tanpa melalui ujian tertulis. Penelitian ini bertujuan mengklasifikasikan mahasiswa program sarjana IPB yang diterima melalui jalur SNBP berdasarkan kategori IP semester 1 menggunakan metode XGBoost dengan penanganan ketidakseimbangan data menggunakan pendekatan EasyEnsemble. Pada metode XGBoost dilakukan hyperparameter tuning dengan pendekatan Fractional Factorial Design 34-1. Data IP dikategorikan ke dalam skema tiga kelas dan empat kelas, kemudian dievaluasi menggunakan K-Fold Cross Validation. Kinerja model diukur menggunakan metrik akurasi, G-Mean, presisi, sensitivitas, dan skor F1, sedangkan feature importance berdasarkan nilai gain XGBoost digunakan untuk mengidentifikasi peubah yang berpengaruh. Hasil penelitian menunjukkan bahwa model XGBoost dengan EasyEnsemble pada skema tiga kelas memberikan performa klasifikasi yang paliing seimbang dengan nilai G-Mean sebesar 58.18% dan sensitivitas sebesar 47.44%, sehingga lebih mampu menangani ketidakseimbangan kelas, terutama pada kelas minoritas. Peubah yang paling berpengaruh meliputi nilai rata-rata keseluruhan, rata-rata nilai mata pelajaran yang menjadi syarat, program studi, provinsi sekolah, dan klaster PPKU. Penerapan EasyEnsemble pada XGBoost dengan hyperparameter tuning menggunakan Fractional Factorial Design 3^4-1 menghasilkan model klasifikasi terbaik pada skema tiga kelas.
       
      The implementation of the Merdeka Curriculum emphasizes flexible and competency-based learning, resulting in students with a more diverse range of academic backgrounds. This is closely related to the SNBP admission scheme, which selects students based on academic performance without a written examination. This study aims to classify undergraduate students at IPB admitted through the SNBP based on their first-semester GPA categories using XGBoost, with EasyEnsemble employed to address class imbalance. Hyperparameter tuning of XGBoost was conducted using a 34-1 Fractional Factorial Design (FFD). GPA data are categorized into three-class and four-class schemes and evaluated using K-Fold Cross Validation. Model performance is measured using accuracy, G-Mean, precision, recall, and F1-score, while feature importance is assessed using XGBoost gain values to identify influential variables. The results showed that the three-class XGBoost model with EasyEnsemble achieves the most balanced classification performance, with a G-Mean of 58.18% and a sensitifity of 47.44%. These result indicate that the proposed model was more effective in handling class imbalance, particularly for the minority classes. The most influential variables include overall average grades, average required grades, study program, school province, and “PPKU” cluster, which directly affect the model’s results and first-semester GPA. The implementation of EasyEnsemble on XGBoost with hyperparameter tuning using a 3^4-1 FFD resulted in the best-performing classification model for the three-class scheme.
       
      URI
      http://repository.ipb.ac.id/handle/123456789/174613
      Collections
      • UF - Statistics and Data Sciences [166]

      Copyright © 2020 Library of IPB University
      All rights reserved
      Contact Us | Send Feedback
      Indonesia DSpace Group 
      IPB University Scientific Repository
      UIN Syarif Hidayatullah Institutional Repository
      Universitas Jember Digital Repository
        

       

      Browse

      All of IPB RepositoryCollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

      My Account

      Login

      Application

      google store

      Copyright © 2020 Library of IPB University
      All rights reserved
      Contact Us | Send Feedback
      Indonesia DSpace Group 
      IPB University Scientific Repository
      UIN Syarif Hidayatullah Institutional Repository
      Universitas Jember Digital Repository