@misc{eprints87234,
           month = {Juni},
           title = {IMPLEMENTASI MODEL BAYESIAN STRUCTURAL TIME SERIES

TERHADAP PREDIKSI NILAI EKSPOR IMPOR MIGAS DAN NON-
MIGAS},
          author = { Annisa R.S	 Claudya},
         address = {UNIVERSITAS LAMPUNG},
       publisher = {FAKULTAS MATEMATIKA DAN ILMU PENGETAHUAN ALAM},
            year = {2024},
             url = {http://digilib.unila.ac.id/87234/},
        abstract = {Model Bayesian Structural Time Series (BSTS) adalah salah satu model yang
dapat digunakan untuk peramalan dengan mempertimbangkan variabel
independen. Pada model BSTS , algoritma Markov Chain Monte Carlo (MCMC)
digunakan untuk mensimulasikan distribusi posterior yang menghaluskan hasil
peramalan lebih akurat dengan menggabungkan hasil dari sejumlah besar model
potensial melalui rata-rata model Bayesian. Penelitian ini bertujuan untuk
mengevaluasi penggunaan model BSTS dalam peramalan data Ekspor Impor
Migas dan Non-Migas di Indonesia dari Januari 2009 hingga Desember 2023.
Penelitian ini memfokuskan pada identifikasi model BSTS terbaik berdasarkan
nilai R-squared, dengan menggunakan komponen tren, musiman, dan regresi serta
mengidentifikasi variabel-variabel yang signifikan memengaruhi nilai Ekspor
Impor Migas dan Non-Migas menggunakan nilai posterior inclusion probability.
Hasil analisis menunjukkan bahwa model BSTS terbaik mencakup komponen
state tren linear lokal dan musiman, tren linear semi lokal dan musiman, serta level
lokal dan musiman dengan jumlah musim yang tepat dan iterasi MCMC yang
optimal.

Kata kunci: model Bayesian Structural Time Series, Markov Chain Monte Carlo,
Prior Spike dan Slab

Bayesian Structural Time Series (BSTS) model is one of the models that can be
used for forecasting while considering independent variables. In BSTS model,
Markov Chain Monte Carlo (MCMC) algorithm is employed to simulate the
posterior distribution, which smoothens the forecasted results more accurately by
combining results from numerous potential models through Bayesian model
averaging. This study aims to evaluate the utilization of the BSTS model in
forecasting the data of Oil and Non-Oil Exports and Imports in Indonesia from
January 2009 to December 2023. The research focuses on identifying the best
BSTS model based on the R-squared value, utilizing components such as trends,
seasonality, and regression, and identifying significant variables that affect Oil and
Non-Oil Exports and Imports using the posterior inclusion probability. The
analysis results indicate that the best BSTS model includes local linear trend and
seasonal state components, semi-local linear trend and seasonal components, as
well as local level and seasonal components with the appropriate number of
seasons and optimal MCMC iterations.
Keywords: Bayesian Structural Time Series Model, Markov Chain Monte Carlo,
Prior Spike dan Slab.}
}