PENERAPAN MULTIVARIATE SINGULAR SPECTRUM ANALYSIS PADA PREDIKSI CUACA DI STASIUN METEOROLOGI PERAK I SURABAYA

Mukti, Abdul (2025) PENERAPAN MULTIVARIATE SINGULAR SPECTRUM ANALYSIS PADA PREDIKSI CUACA DI STASIUN METEOROLOGI PERAK I SURABAYA. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text COVER (1).pdf Download (958kB)
	Text BAB I.pdf Download (222kB)
	Text BAB II.pdf Restricted to Repository staff only until 17 September 2028. Download (621kB)
	Text BAB III.pdf Restricted to Repository staff only until 17 September 2028. Download (552kB)
	Text BAB IV.pdf Restricted to Repository staff only until 17 September 2028. Download (1MB)
	Text BAB V.pdf Download (153kB)
	Text DAFTAR PUSTAKA.pdf Download (160kB)
	Text LAMPIRAN.pdf Restricted to Repository staff only until 17 September 2028. Download (161kB)

Abstract

Weather has a significant impact on various aspects of human life, such as agriculture, transportation, and health. Unexpected weather changes, especially in Indonesia which is vulnerable to extreme phenomena, demand accurate prediction methods. This study aims to predict weather in Surabaya using the Multivariate Singular Spectrum Analysis (MSSA) method. This method has advantages because it can analyze multivariate data, is non-parametric, and does not require assumptions of stationarity, normality, or linearity. Weather data is time series and multivariate with complex patterns influenced by long-term trends, seasonal fluctuations, and irregular extreme events. Therefore, the use of MSSA is relevant to capture weather dynamics more comprehensively through decomposition and reconstruction processes. The research data consists of daily records of minimum temperature (TN), maximum temperature (TX), average temperature (TAVG), and average humidity (RH_AVG) from the Meteorological Station Perak 1 Surabaya for the period August 1, 2024 to January 7, 2025. The analysis was carried out through a decomposition stage to generate matrices and eigentriples, and a reconstruction stage to rebuild the time series. The results show a window length (L) of 76 with three main groups representing trend and seasonal components. Model performance evaluation in the first modeling (with RH_AVG) produced MAPE values of 3.31% (TN), 5.01% (TX), 4.52% (TAVG), and 6.79% (RH_AVG). Meanwhile, the second modeling (without RH_AVG) showed improved accuracy with lower MAPE values, namely 1.91% (TN), 2.95% (TX), and 2.63% (TAVG). These results prove that eliminating negatively correlated variables contributes to improving model performance.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Hindrayani, Kartika Maulida	NIDN0009099205	kartika.maulida.ds@upnjatim.ac.id
Thesis advisor	Idhom, Mohammad	NIDN0010038305	idhom@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Data Science

Depositing User:

Abdul Mukti

Date Deposited:

19 Sep 2025 03:29

Last Modified:

19 Sep 2025 03:35

URI:

https://repository.upnjatim.ac.id/id/eprint/43689