PERAMALAN NILAI TUKAR PETANI MENGGUNAKAN METODE HYBRID AUTOREGERESSIVE INTEGRATED MOVING AVERAGE-ARTIFICIAL NEURAL NETWORK (ARIMA-ANN) DI PROVINSI JAMBI

Panjaitan, Kevin (2025) PERAMALAN NILAI TUKAR PETANI MENGGUNAKAN METODE HYBRID AUTOREGERESSIVE INTEGRATED MOVING AVERAGE-ARTIFICIAL NEURAL NETWORK (ARIMA-ANN) DI PROVINSI JAMBI. S1 thesis, UNIVERSITAS JAMBI.

	Text ABSTRAK.pdf Download (53kB)
	Text COVER.pdf Download (121kB)
	Text BAB I.pdf Download (39kB)
	Text BAB V.pdf Download (27kB)
	Text DAFTAR PUSTAKA.pdf Download (50kB)
	Text LEMBAR PENGESAHAN.pdf Download (111kB)
	Text FULL SKRIPSI.pdf Restricted to Repository staff only Download (1MB)

Official URL: https://repository.unja.ac.id/

Abstract

The agricultural sector absorbs 29.36% of the national workforce and contributes exports worth 374 million USD. In Jambi Province, 45.44% of the population works in this sector, making farmers’ welfare a top priority. One of the main indicators of this welfare is the Farmers’ Exchange Rate (NTP). NTP is a time series dataset compiled on a monthly basis, used to monitor the development of farmers’ welfare over time. To support this monitoring process, a forecasting method is required to predict future NTP values. The Autoregressive Integrated Moving Average (ARIMA) method is effective in capturing linear patterns, while the Artificial Neural Network (ANN) can model nonlinear components indicated by ARIMA residuals. Therefore, the Hybrid ARIMA–ANN approach is employed to combine the strengths of both methods and has been proven to produce more accurate forecasts than single models. The data analysis process in this study began with descriptive data collection and organization to understand the initial characteristics and identify existing patterns. The dataset was divided into 80% training data and 20% testing data. Stationarity tests were performed on the training data, and if non-stationary, transformations and differencing were applied until stationarity was achieved. Next, an ARIMA model was constructed, followed by parameter estimation, diagnostic testing, and validation using the testing data. The residuals from the ARIMA model were then used to build an Artificial Neural Network (ANN) model, which, after validation, was combined into a Hybrid ARIMA–ANN model. This hybrid model was ultimately used to conduct comprehensive forecasting. Based on the results, the forecasting model obtained was Hybrid ARI- MA(5,1,0)(ANN(32,1)), expressed as follows: ˆYt = ˆLt + ˆNt, ˆLt = Yt−1 + 0.2149 (Yt−1 − Yt−2) − 0.1235 (Yt−5 − Yt−6) + 3.5676, ˆNt = 32X j=1 vj ReLU 3X i=1 wj,iet−i + bj ! + c. Forecasting for the next ten periods (April 2025 to January 2026) ranges between 162.52 and 169.98. Thus, it can be concluded that the Hybrid ARIMA–ANN model is effective and accurate in forecasting the Farmers’ Exchange Rate (NTP), as it can simultaneously capture both linear and nonlinear patterns.

Type:	Thesis (S1)
Uncontrolled Keywords:	Nilai Tukar Petani, Peramalan, Hybrid ARIMA–Neural Network.
Subjects:	L Education > L Education (General)
Divisions:	Fakultas Sains dan Teknologi > Matematika
Depositing User:	KEVIN SYNAGOGUE PANJAITAN
Date Deposited:	15 Oct 2025 08:07
Last Modified:	15 Oct 2025 08:07
URI:	https://repository.unja.ac.id/id/eprint/86819

Actions (login required)

View Item