Peramalan Beban Listrik Jangka Pendek Dengan Pola Multiple Seasonal Menggunakan Model Hybrid TSR, DHR, STL, ETS, dan TBATS

Pertiwi, Ilalang Akar (2021) Peramalan Beban Listrik Jangka Pendek Dengan Pola Multiple Seasonal Menggunakan Model Hybrid TSR, DHR, STL, ETS, dan TBATS. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

Text 06211740000004-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (7MB) | Request a copy

Abstract

Listrik merupakan sumber energi yang sangat penting bagi manusia. Permintaan listrik tergantung pada tingkat konsumsi masyarakat di suatu wilayah. Permintaan listrik ini harus diimbangi dengan ketersediaan pasokan listrik yang memadai. Oleh karena itu, peramalan konsumsi listrik menjadi hal yang sangat penting untuk dilakukan. Data yang digunakan dalam penelitian ini merupakan data konsumsi listrik per jam di Turki tahun 2016-2020. Data ini memiliki pola multiple seasonal yang terdiri dari pola musiman harian (1x24 jam), pola musiman mingguan (7x24 jam), dan pola musiman tahunan (365x24 jam), serta variasi kalender yang diakibatkan karena adanya perbedaan waktu antara kalender Masehi dan kalender Hijriah. Kombinasi pemodelan multiple seasonal dan variasi kalender ini dilakukan dengan pendekatan hibrida, yaitu dengan kombinasi model TSR-DHR, TSR-STL+ETS, dan TSR-TBATS. Dan diperoleh bahwa peramalan terbaik untuk data out-sample dengan variasi kalender hari raya yaitu menggunakan model hibrida TSR dan STL+ETS. Sedangkan pada data out-sample tanpa variasi kalender hari raya, model terbaik yang diperoleh yaitu model hibrida TSR dan TBATS. Pendekatan hibrida terbukti meningkatkan akurasi peramalan.
=========================================================================================================
Electricity is a very essential energy source for human activities. The demand for electricity depends on the level of community consumption in an area. This electricity demand must be balanced with the availability of adequate electricity supply. Therefore, forecasting electricity consumption is a very important thing to do. The data used in this study is data on electricity consumption per hour in Turkey in 2016-2020. This data has multiple seasonal patterns consisting of daily seasonal patterns (1x24 hours), weekly seasonal patterns (7x24 hours), and annual seasonal patterns (365x24 hours), as well as calendar variations caused by time differences between the Gregorian calendar and the Hijri calendar. The combination of multiple seasonal and calendar variations modelling is carried out using a hybrid approach, that are combination of TSR-DHR, TSR-STL+ETS, and TSR-TBATS models. And it is found that the best forecasting model for out-sample data with calendar variation in the form of Muslim holidays is by using the TSR and STL+ETS hybrid model. While the best model obtained for the out-sample data without calendar variation of the Muslim holiday, is the TSR and TBATS hybrid model. A hybrid approach is proven to improve forecasting accuracy.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Complex Seasonality, Dynamic Regression, Model Hibrida, STLF, Variasi Kalender, Calendar Variation, Hybrid Model
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA280 Box-Jenkins forecasting Q Science > QA Mathematics > QA404 Fourier series T Technology > T Technology (General) > T174 Technological forecasting
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Ilalang Akar Pertiwi
Date Deposited:	01 Sep 2021 15:21
Last Modified:	01 Sep 2021 15:21
URI:	http://repository.its.ac.id/id/eprint/91448

Actions (login required)

View Item