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

      Perbandingan Performa Model LSTM dan GRU pada Data Deret Waktu Multivariat dalam Peramalan Indeks Harga Saham Gabungan

      Thumbnail
      View/Open
      Cover (676.9Kb)
      Fulltext (2.139Mb)
      Lampiran (2.338Mb)
      Date
      2025
      Author
      Radithya, Rafli
      Sadik, Kusman
      Dito, Gerry Alfa
      Metadata
      Show full item record
      Abstract
      Penelitian ini membandingkan kinerja dua arsitektur deep learning, yaitu long short-term memory (LSTM) dan gated recurrent unit (GRU), dalam melakukan peramalan multivariat multi-step terhadap Indeks Harga Saham Gabungan (IHSG) menggunakan pendekatan multi-input multi-output (MIMO). Data historis periode 31 Juli 2000 hingga 8 Mei 2025, yang terdiri atas nilai IHSG dan 12 peubah eksternal, digunakan untuk memprediksi pergerakan IHSG pada horizon satu hingga lima hari. Validasi dilakukan menggunakan skema expanding window. Hasil penelitian menunjukkan bahwa kedua model menghasilkan akurasi prediksi yang baik, namun model GRU terbaik secara konsisten memberikan kinerja lebih unggul pada semua horizon. Model GRU terbaik mencapai nilai rata-rata MAPE 4,28% dan RMSE sebesar 235,42. Pada horizon prediksi satu hingga lima hari, nilai MAPE masing-masing adalah 10,715%, 5,885%, 1,748%, 1,847%, dan 1,216%. Sementara itu, nilai RMSE untuk horizon prediksi hari pertama hingga kelima berturut-turut adalah 467,309; 327,159; 125,928; 143,453; dan 113,265. Maka, model terbaik GRU dipilih sebagai model terbaik. Analisis permutation feature importance mengidentifikasi bahwa nilai IHSG satu hingga tiga hari sebelumnya merupakan peubah paling dominan, mencerminkan auto korelasi kuat pada deret waktu IHSG. Peubah lain yang berpengaruh signifikan adalah indeks saham kawasan Asia Tenggara dan harga emas.
       
      This study compares the forecasting performance of two deep learning architectures—long short-term memory (LSTM) and gated recurrent unit (GRU)—for multivariate, multi-step prediction of the Jakarta Stock Exchange (JKSE) index using the multi-input multi-output (MIMO) approach. Historical data from 31 July 2000 to 8 May 2025, consisting of the JKSE index and twelve external predictors, were used to generate 1–5 day ahead forecasts. Model evaluation was conducted using an expanding-window validation strategy. The results indicate that both LSTM and GRU best models achieve strong predictive accuracy, with the GRU consistently outperforming LSTM across all forecasting horizons. The best GRU model achieved an average MAPE of 4.28% and an RMSE of 235.4229. For one- to five-day forecasting horizons, the MAPE values were 10,715%, 5,885%, 1,748%, 1,847%, and 1,216%, respectively. The corresponding RMSE values were 467,309; 327,159; 125,928; 143,453; and 113,265. Based on these results, making it the optimal model for this study. Permutation feature importance analysis reveals that the current JKSE value and its lags up to three days are the most influential predictors, reflecting strong autocorrelation patterns in the data. Southeast Asian stock indices and gold prices also contribute substantially to the model’s predictive performance.
       
      URI
      http://repository.ipb.ac.id/handle/123456789/171816
      Collections
      • UT - Statistics and Data Sciences [85]

      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