Penerapan Holt-Winters untuk Peramalan Harga Beras di Provinsi Jawa Timur dengan Pendekatan Time Series

Isnaini, Frisda Dita (2024) Penerapan Holt-Winters untuk Peramalan Harga Beras di Provinsi Jawa Timur dengan Pendekatan Time Series. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 20081010193.-cover.pdf Download (525kB)
	Text (Bab 1) 20081010193.-bab1.pdf Download (117kB)
	Text (Bab 2) 20081010193.-bab2.pdf Restricted to Repository staff only until 24 July 2026. Download (577kB)
	Text (Bab 3) 20081010193.-bab3.pdf Restricted to Repository staff only until 24 July 2026. Download (902kB)
	Text (Bab 4) 20081010193.-bab4.pdf Restricted to Repository staff only until 24 July 2026. Download (11MB)
	Text (Bab 5) 20081010193.-bab5.pdf Download (96kB)
	Text (Daftar Pustaka) 20081010193.-daftarpustaka.pdf Download (165kB)

Abstract

Rice is a staple food consumed by the majority of the Indonesian population. As an agrarian country, Indonesia had 15,550,786 agricultural businesses by 2023, with paddy rice being the most widely cultivated commodity. However, rice prices have risen significantly, reaching Rp19,550 in 2024 across all regions of Indonesia, including East Java, the largest rice producer, currently experiencing a period of scarcity. This situation raises questions about the price movement patterns of rice in Indonesia, particularly in East Java, over a certain period. Forecasting rice prices enables policymakers, farmers, and other stakeholders to take appropriate actions regarding production, distribution, and policy-making to maintain rice price stability. This study aims to model the forecasting of rice prices in 20 regions of East Java from 2017 to 2023 using Holt-Winters Exponential Smoothing. The Holt-Winters Exponential Smoothing model emphasizes forecasting on the level, trend, and seasonal components. The results indicate that rice prices tend to rise during the year-end or from the end to the beginning of the year and start to decline mid-year. The study tested the model by splitting the data using K-Fold cross-validation with k values of 3 and 5, and tested the range of alpha, beta, and gamma parameters from 0.1 to 0.9 and 0.01 to 0.9. The testing results show that the optimal parameters are alpha 0.9, beta 0.01, and gamma 0.9 with k = 5. The model testing process yielded the best MAPE error value in Banyuwangi with a percentage of 0.03%.

Item Type:

Thesis (Undergraduate)

Contributors: