Peramalan Temperatur Bandung 2024–2025 dengan Metode Multivariate Singular Spectrum Analysis

Abstract. Temperature changes in Bandung City have shown a fluctuating trend. In recent times, the temperature has tended to increase due to a prolonged dry season. In fact, in early October 2023, the temperature in Bandung reached 36°C. Therefore, an accurate forecasting method is needed to understand temperature change patterns in order to support better planning. This study applies the Multivariate Singular Spectrum Analysis (MSSA) method to forecast minimum, average, and maximum temperatures in Bandung City from 2016 to 2023.MSSA is an extension of the univariate SSA forecasting method, developed to handle multivariate time series by incorporating covariances among variables when calculating weights in the forecasting model. This enables the simultaneous forecasting of several time series variables. MSSA is a nonparametric statistical method, as it does not require assumptions such as stationarity or normality of the data.The MSSA analysis process includes decomposition through embedding and Singular Value Decomposition (SVD), reconstruction through grouping and diagonal averaging, and forecasting using the Linear Recurrent Formula (LRF). The forecasting results are then evaluated using the Mean Absolute Percentage Error (MAPE) to measure the accuracy of the model.This research is expected to serve as a reference for climate data analysis and forecasting, as well as support policy formulation related to temperature changes in urban areas. The results indicate that MSSA yields a low MAPE value, signifying its strong performance in forecasting air temperature in Bandung. Thus, MSSA can be considered an effective alternative for multivariate forecasting, especially for complex and non-stationary data patterns. Abstrak. Perubahan temperatur di Kota Bandung menunjukkan tren yang fluktuatif,dalam beberapa waktu terakhir, suhu di Bandung cenderung lebih panas akibat musim kemarau yang panjang. Bahkan, pada awal Oktober 2023, suhu di Bandung sempat mencapai 36 derajat Celcius. Oleh karena itu, diperlukan metode peramalan yang akurat untuk memahami pola perubahan suhu guna mendukung perencanaan yang lebih baik. Penelitian ini menerapkan Multivariate Singular Spectrum Analysis (MSSA) untuk meramalkan temperatur minimal, rata-rata, dan maksimal di Kota Bandung dari tahun 2016 hingga 2023. Metode ini dikembangkan dari metode peramalan univariat SSA dalam menangani data multivariat, dengan cara melibatkan kovarians diantara variabel didalam mehitung bobot-bobot dalam model peramalan, sehingga beberapa variabel deret waktu, nilainya dapat diramalkan secara simultan. Metode MSSA termasuk pada statistika nonparametrik karena tidak mengasumsikan seperti kestationeran dan kenormalan data. Proses analisis MSSA melibatkan tahapan dekomposisi dengan Embedding dan Singular Value Decomposition (SVD), rekonstruksi dengan grouping dan diagonal averaging, serta peramalan menggunakan Linear Recurrent Formula (LRF). Hasil peramalan kemudian dievaluasi dengan Mean Absolute Percentage Error (MAPE) untuk mengukur tingkat akurasi model. Diharapkan penelitian ini dapat menjadi referensi dalam analisis dan peramalan data iklim serta mendukung perumusan kebijakan terkait perubahan suhu di wilayah perkotaan. Hasil penelitian menunjukkan bahwa metode MSSA memberikan nilai MAPE yang rendah, dimana ini mengindikasikan bahwa MSSA unggul dalam meramalkan suhu udara di Kota Bandung. Dengan demikian, MSSA dapat dijadikan alternatif yang lebih efektif dalam analisis peramalan multivariat, terutama untuk data dengan pola kompleks dan non-stasioner.

Authors

Institutions

Publication Details

Journal
Bandung Conference Series Statistics
Published
2025-07-12
DOI
https://doi.org/10.29313/bcss.v5i2.18420
Primary Topic
Management and Optimization Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Peramalan Temperatur Bandung 2024–2025 dengan Metode Multivariate Singular Spectrum Analysis

Suwanda, Kariim Kurniawan Pamungkas Andawali Putra
Bandung Conference Series Statistics
Management and Optimization Techniques
article

Peramalan Temperatur Bandung 2024–2025 dengan Metode Multivariate Singular Spectrum Analysis

Suwanda, Kariim Kurniawan Pamungkas Andawali Putra
article en

Abstract

Abstract. Temperature changes in Bandung City have shown a fluctuating trend. In recent times, the temperature has tended to increase due to a prolonged dry season. In fact, in early October 2023, the temperature in Bandung reached 36°C. Therefore, an accurate forecasting method is needed to understand temperature change patterns in order to support better planning. This study applies the Multivariate Singular Spectrum Analysis (MSSA) method to forecast minimum, average, and maximum temperatures in Bandung City from 2016 to 2023.MSSA is an extension of the univariate SSA forecasting method, developed to handle multivariate time series by incorporating covariances among variables when calculating weights in the forecasting model. This enables the simultaneous forecasting of several time series variables. MSSA is a nonparametric statistical method, as it does not require assumptions such as stationarity or normality of the data.The MSSA analysis process includes decomposition through embedding and Singular Value Decomposition (SVD), reconstruction through grouping and diagonal averaging, and forecasting using the Linear Recurrent Formula (LRF). The forecasting results are then evaluated using the Mean Absolute Percentage Error (MAPE) to measure the accuracy of the model.This research is expected to serve as a reference for climate data analysis and forecasting, as well as support policy formulation related to temperature changes in urban areas. The results indicate that MSSA yields a low MAPE value, signifying its strong performance in forecasting air temperature in Bandung. Thus, MSSA can be considered an effective alternative for multivariate forecasting, especially for complex and non-stationary data patterns. Abstrak. Perubahan temperatur di Kota Bandung menunjukkan tren yang fluktuatif,dalam beberapa waktu terakhir, suhu di Bandung cenderung lebih panas akibat musim kemarau yang panjang. Bahkan, pada awal Oktober 2023, suhu di Bandung sempat mencapai 36 derajat Celcius. Oleh karena itu, diperlukan metode peramalan yang akurat untuk memahami pola perubahan suhu guna mendukung perencanaan yang lebih baik. Penelitian ini menerapkan Multivariate Singular Spectrum Analysis (MSSA) untuk meramalkan temperatur minimal, rata-rata, dan maksimal di Kota Bandung dari tahun 2016 hingga 2023. Metode ini dikembangkan dari metode peramalan univariat SSA dalam menangani data multivariat, dengan cara melibatkan kovarians diantara variabel didalam mehitung bobot-bobot dalam model peramalan, sehingga beberapa variabel deret waktu, nilainya dapat diramalkan secara simultan. Metode MSSA termasuk pada statistika nonparametrik karena tidak mengasumsikan seperti kestationeran dan kenormalan data. Proses analisis MSSA melibatkan tahapan dekomposisi dengan Embedding dan Singular Value Decomposition (SVD), rekonstruksi dengan grouping dan diagonal averaging, serta peramalan menggunakan Linear Recurrent Formula (LRF). Hasil peramalan kemudian dievaluasi dengan Mean Absolute Percentage Error (MAPE) untuk mengukur tingkat akurasi model. Diharapkan penelitian ini dapat menjadi referensi dalam analisis dan peramalan data iklim serta mendukung perumusan kebijakan terkait perubahan suhu di wilayah perkotaan. Hasil penelitian menunjukkan bahwa metode MSSA memberikan nilai MAPE yang rendah, dimana ini mengindikasikan bahwa MSSA unggul dalam meramalkan suhu udara di Kota Bandung. Dengan demikian, MSSA dapat dijadikan alternatif yang lebih efektif dalam analisis peramalan multivariat, terutama untuk data dengan pola kompleks dan non-stasioner.

Bandung Conference Series StatisticsVol. 5(2)
Bandung Islamic University (ID)
Openalex Percentile: Top 68%
Management and Optimization Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.