PERAMALAN CURAH HUJAN KOTA BANDUNG DENGAN MODEL HYBRID SARIMAX-LSTM

Dewi Qurotul Aeni, - (2024) PERAMALAN CURAH HUJAN KOTA BANDUNG DENGAN MODEL HYBRID SARIMAX-LSTM. S1 thesis, Universitas Pendidikan Indonesia.

	Text S_MAT_2008140_Title.pdf Download (630kB)
	Text S_MAT_2008140_Chapter1.pdf Download (95kB)
	Text S_MAT_2008140_Chapter2.pdf Restricted to Staf Perpustakaan Download (443kB)
	Text S_MAT_2008140_Chapter3.pdf Download (223kB)
	Text S_MAT_2008140_Chapter4.pdf Restricted to Staf Perpustakaan Download (939kB)
	Text S_MAT_2008140_Chapter5.pdf Download (101kB)
	Text S_MAT_2008140_Appendix.pdf Restricted to Staf Perpustakaan Download (965kB)

Abstract

Curah hujan merupakan fenomena alam yang memiliki peran krusial dalam kehidupan manusia dan ekosistem di dunia. Prediksi curah hujan yang akurat memainkan peran penting dalam mengelola sumber daya alam serta dalam upaya mitigasi dan perencanaan yang lebih baik terkait pengelolaan air, keselamatan masyarakat, serta pembangunan infrastruktur. Model yang digunakan dalam peramalan curah hujan dalam penelitian ini adalah model hybrid SARIMAX-LSTM (Seasonal Autoregressive Integrated Moving Average with Exogenous Variable – Long Short Term Memory). Model SARIMAX adalah model SARIMA dengan tambahan variabel eksogen, dalam penelitian ini berupa nilai persentase kelembaban udara dalam periode waktu yang sama. Selain itu, model LSTM digunakan untuk menutupi kekurangan model SARIMAX, yaitu kurang mampu menangani peramalan jangka panjang dan non-linear secara efisien. Penelitian dimulai dengan menjalankan model SARIMAX, kemudian residu dari hasil model tersebut diolah dengan model LSTM. Model terbaik yang diperoleh adalah model SARIMAX (2,1,0)(2,1,0)(12) dan model LSTM yang menggunakan fungsi aktivasi Relu dengan jumlah window size 12, batch size sebanyak 32, dan memiliki 2 hidden layer dengan parameter optimal sebanyak 128 neurons serta melalui proses pelatihan sejumlah 25 epochs. Hasil akhir peramalan curah hujan Kota Bandung menunjukkan bahwa model hybrid SARIMAX-LSTM lebih baik dibandingkan model SARIMAX, dengan curah hujan tertinggi terjadi pada bulan Maret sebesar 375,291 mm dan curah hujan terendah terjadi pada bulan September sebesar 57,535 mm. Rainfall is a natural phenomenon that plays a crucial role in human life and ecosystems around the world. Accurate rainfall prediction plays an important role in managing natural resources as well as in mitigation efforts and better planning related to water management, public safety, and infrastructure development. The model used in rainfall forecasting in this study is the hybrid SARIMAX-LSTM (Seasonal Autoregressive Integrated Moving Average with Exogenous Variable - Long Short Term Memory) model. The SARIMAX model is a SARIMA model with additional exogenous variables, in this study in the form of the percentage value of air humidity in the same time period. In addition, the LSTM model is used to cover the shortcomings of the SARIMAX model, which is less able to handle long-term and non-linear forecasting efficiently. The research begins by running the SARIMAX model, then the residuals from the model results are processed with the LSTM model. The best model obtained is the SARIMAX (2,1,0)(2,1,0)(12) model and the LSTM model which uses the Relu activation function with a window size of 12, batch size of 32, and has 2 hidden layers with optimal parameters of 128 neurons and through a training process of 25 epochs. The final results of Bandung City rainfall forecasting show that the hybrid SARIMAX-LSTM model is better than the SARIMAX model, with the highest rainfall occurring in March at 375.291 mm and the lowest rainfall occurring in September at 57.535 mm.

Item Type:	Thesis (S1)
Additional Information:	https://scholar.google.com/citations?view_op=new_profile&hl=en ID SINTA Dosen Pembimbing: Fitriani Agustina: 5981275 Lukman: 6675529
Uncontrolled Keywords:	Runtun waktu, Curah hujan, SARIMAX, LSTM, SARIMAX-LSTM. Time series, Precipitation, SARIMAX, LSTM, SARIMAX-LSTM.
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions:	Fakultas Pendidikan Matematika dan Ilmu Pengetahuan Alam > Jurusan Pendidikan Matematika > Program Studi Matematika (non kependidikan)
Depositing User:	Dewi Qurotul Aeni
Date Deposited:	09 Sep 2024 03:22
Last Modified:	09 Sep 2024 03:22
URI:	http://repository.upi.edu/id/eprint/122846

Actions (login required)

View Item