Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/170535
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorAfendi, Farit Mochamad-
dc.contributor.advisorMasjkur, Mohammad-
dc.contributor.authorZulfikar, Muhammad Raziv-
dc.date.accessioned2025-08-27T03:55:56Z-
dc.date.available2025-08-27T03:55:56Z-
dc.date.issued2025-
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/170535-
dc.description.abstractTelur ayam merupakan salah satu sumber proten hewani yang memiliki gizi tinggi, harga terjangkau dan mudah diakses masyarakat. Karakteristik tersebut membuat telur ayam memiliki peran strategis dalam menjaga ketahanan pangan dan gizi masyarakat Indonesia, terutama untuk masyarakat berpenghasilan rendah. Penelitian ini bertujuan membandingkan kinerja model deep learning sekuensial seperti Long Short-Term Memory (LSTM) dan Gated Recurrent Unit (GRU) dengan model paralel seperti Autoformer dan Informer dalam memprediksi harga telur harian di Jawa Barat. Pada penelitian ini digunakan dua pendekatan pemodelan yaitu, model single-output dan multi-output. Hasil penelitian menunjukkan bahwa model Informer lebih unggul dibandingkan model lainnya pada kedua pendekatan pemodelan, dengan nilai RMSE terendah dan paling stabil. Model sekuensial seperti GRU dan LSTM memiliki waktu latih model yang lebih cepat dibandingkan model paralel. Penelitian ini juga menerapkan model terbaik dari masing-masing pendekatan untuk meramal harga telur pada sebulan ke depan. Model single-output memprediksi bahwa harga telur akan mengalami penurunan signifikan, diikuti oleh kenaikan harga pada akhir periode peramalan. Sementara itu, model multi-output memprediksi harga telur akan menunjukkan tren yang lebih stabil dengan hasil yang lebih mendekati data aktual yang sudah tersedia.-
dc.description.abstractChicken eggs are one of the primary sources of animal protein, which are recognized as being nutritious, affordable, and readily available. These characteristics make eggs a vital component in supporting food and nutrition security, particularly for low and middle-income households. The objective of this study is to make a comparison between the performance of sequential deep learning models, such as Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU), and parallel models, such as Autoformer and Informer, on predicting daily egg prices in West Java. In this study, two modeling approach will be applied, single-output and multi-output model. The results of the study show that the Informer model outperforms the others in both modeling approach, achieving the lowest and most stable RMSE values. However, in terms of training time, sequential models like GRU and LSTM have shorter training times compared to parallel models. This study also applied the best models from each modeling approach to predict future egg prices. The single-output model predicted that egg prices would experience a significant decline over the next month, followed by a gradual price increase. Meanwhile, the multi-output model predicted that egg prices more stable trend over next, closely resembling the actual observed data.-
dc.description.sponsorshipnull-
dc.language.isoid-
dc.publisherIPB Universityid
dc.titlePerbandingan Model Time Series Berbasis Deep Learning dalam Memprediksi Harga Telur: GRU, LSTM, Informer dan Autoformerid
dc.title.alternativenull-
dc.typeSkripsi-
dc.subject.keywordLong Short-Term Memory (LSTM)id
dc.subject.keywordharga telur ayamid
dc.subject.keywordGated Recurrent Unit (GRU)id
dc.subject.keywordInformerid
dc.subject.keywordAutoformerid
Appears in Collections:UT - Statistics and Data Sciences

Files in This Item:
File Description SizeFormat 
cover_G1401201083_231a379f5b76410a90c8b45e94f367d1.pdfCover283.87 kBAdobe PDFView/Open
fulltext_G1401201083_c96d6be12b3d490f87a5f7ee34d9bb7c.pdf
  Restricted Access
Fulltext1.72 MBAdobe PDFView/Open
lampiran_G1401201083_d86db9b6b1704850818bfb9ee0239c3d.pdf
  Restricted Access
Lampiran180.79 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.