Implementasi Model GSTAR (Generalized Space Time Autoregressive) dan GSTAR-Kalman Filter dalam Peramalan Harga Cabai Rawit Merah

Lestari, Indah Dwi Aprilia (2024) Implementasi Model GSTAR (Generalized Space Time Autoregressive) dan GSTAR-Kalman Filter dalam Peramalan Harga Cabai Rawit Merah. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002201028-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Provinsi Jawa Timur memiliki berbagai macam komoditas pangan unggulan, salah satunya adalah cabai. Kebutuhan cabai juga sangat tinggi sehingga permintaan cabai khususnya di Jawa Timur sangat besar dan hal ini menyebabkan harga cabai terus mengalami fluktuasi terutama menjelang hari-hari besar. Peramalan mengenai harga cabai sangat diperlukan bagi pemerintah, industri, dan rumah tangga. Peramalan harga cabai termasuk ke dalam jenis peramalan data spasial-temporal karena memiliki hubungan antar waktu dan lokasi. Salah satu model peramalan untuk data spasialtemporal adalah model GSTAR. Model GSTAR yang digunakan memiliki keterbatasan yaitu hanya menggunakan orde spasial 1, sehingga dibutuhkan metode pemulusan untuk memperkecil nilai error, salah satunya adalah metode Kalman-Filter. Dengan demikian, penulis melakukan penelitian mengenai implementasi model GSTAR dan GSTAR-Kalman Filter dalam peramalan harga cabai rawit merah. Penelitian yang dilakukan menggunakan data sekunder Bulan Februari 2018 sampai Februari 2024 dari empat kota di Jawa Timur yaitu Kota Blitar, Kediri, Probolinggo, dan Malang dengan bobot lokasi invers jarak. Penelitian ini bertujuan untuk mendapatkan model GSTAR dan GSTAR-Kalman Filter untuk meramalkan harga cabai rawit merah di Kota Blitar, Kediri, Probolinggo, dan Malang. Selain itu, penelitian ini juga bertujuan untuk mendapatkan hasil peramalan dan membandingkan keakuratan kedua model tersebut dalam meramalkan harga cabai rawit merah di kota-kota tersebut. Model GSTAR yang terpilih dalam penelitian ini adalah GSTAR(1_1), yang kemudian dikembangkan menjadi model GSTAR-Kalman Filter. Hasil analisis menunjukkan bahwa model GSTAR-Kalman Filter memiliki nilai RMSE dan MAPE yang lebih rendah dibandingkan dengan model GSTAR, dengan RMSE sebesar 860.28 dan MAPE sebesar 1.66%.
====================================================================================================================================
East Java Province has various kinds of superior food commodities, one of which is chili. The need for chilies is also very high so the demand for chilies, especially in East Java, is very large and this causes the price of chilies to continue to fluctuate, especially before big holidays. Forecasting regarding chili prices is very necessary for the government, industry and households. Chili price forecasting is included in the type of spatial-temporal data forecasting because it has a relationship between time and location. One of the forecasting models for spatial-temporal data is the GSTAR model. The GSTAR model used has limitations, namely that it only uses spatial order 1, so a smoothing method is needed to reduce the error value, one of which is the Kalman-Filter method. Thus, the author conducted research regarding the implementation of the GSTAR and GSTAR-Kalman Filter models in forecasting the price of red cayenne pepper. The research was conducted using secondary data from February 2018 to February 2024 from four cities in East Java, namely Blitar, Kediri, Probolinggo and Malang with the location weights used being the inverse distance location weights. The objective of the upcoming research is to develop the GSTAR and GSTAR-Kalman Filter models for forecasting the price of cayenne pepper in the cities of Blitar, Kediri, Probolinggo, and Malang. The research aims to generate forecasting results and compare the effectiveness of the GSTAR and GSTAR-Kalman Filter models in forecasting the price of red cayenne pepper in these cities. The selected GSTAR model is GSTAR(1_1) which is then formed into the GSTAR-Kalman Filter model. Based on these two models, the GSTAR-Kalman Filter model has a smaller RMSE and MAPE compared to the GSTAR model, namely RMSE of 860.28 and MAPE of 1.66%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	GSTAR, Price of Red Cayenne Pepper, Kalman Filter, Forecasting, GSTAR, Harga Cabai Rawit Merah, Kalman Filter, Peramalan
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA402.3 Kalman filtering. Q Science > QA Mathematics > QA278 Cluster Analysis. Multivariate analysis. Correspondence analysis (Statistics)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Indah Dwi Aprilia Lestari
Date Deposited:	25 Jul 2024 06:20
Last Modified:	25 Jul 2024 06:20
URI:	http://repository.its.ac.id/id/eprint/108891

Actions (login required)

View Item