PENERAPAN MODEL HIBRIDA ARIMA-LSTM PADA PREDIKSI INFLASI DI INDONESIA

Ameera, Divanda Shaffa (2025) PENERAPAN MODEL HIBRIDA ARIMA-LSTM PADA PREDIKSI INFLASI DI INDONESIA. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21083010031_Cover.pdf Download (2MB)
	Text (Bab 1) 2103010031_Bab I.pdf Download (575kB)
	Text (Bab 2) 21083010031_Bab II.pdf Restricted to Registered users only until 24 May 2028. Download (658kB)
	Text (Bab 3) 21083010031_Bab III.pdf Restricted to Registered users only until 24 May 2028. Download (89kB)
	Text (Bab 4) 21083010031_Bab IV.pdf Restricted to Registered users only until 24 May 2028. Download (3MB)
	Text (Bab 5) 21083010031_Bab V.pdf Download (7kB)
	Text (Daftar Pustaka) 21083010031_Daftar Pustaka.pdf Download (245kB)
	Text (Lampiran) 21083010031_Lampiran.pdf Restricted to Registered users only until 24 May 2028. Download (123kB)

Abstract

Inflation is a sustained increase in the overall prices of goods and services over a specific period of time. Controlling inflation is essential for maintaining economic stability and public welfare. Therefore, effective planning and management of inflation are key to preserving economic stability. One approach to support this is by using appropriate forecasting methods. Based on this background, this study aims to apply a hybrid ARIMA-LSTM model to inflation data in Indonesia. The dataset used spans from 1979 to 2024, indicating the possibility of heteroskedasticity or nonlinear patterns within the data. ARIMA parameter selection was conducted using two approaches: through ACF and PACF plots, and automatic computational methods. The assumption tests on the ARIMA model did not meet the ideal requirements due to violations of its assumptions; therefore, to improve the reliability of the ARIMA model's statistical outcomes, the residuals were predicted using LSTM. The data processing and modeling using the ARIMA-LSTM hybrid approach took approximately 30 minutes, resulting in the combination of ARIMA (3,1,4) and LSTM with evaluation metrics of MAE = 0.21, MSE = 0.08, and RMSE = 0.29. The model demonstrated good performance without overfitting or underfitting. When compared to actual data, the predicted inflation pattern closely resembled the real trend, although it failed to capture the sharp spike in the final prediction. Keywords: AutoRegressive Integrated Moving Average (ARIMA), Inflation, Long Short Term Memory (LSTM, Prediction.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Damaliana, Aviolla Terza	NIDN002089402	aviolla.terza.sada@upnjatim.ac.id
Thesis advisor	Idhom, Mohammad	NIDN0010038305	idhom@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA76.6 Computer Programming

Divisions:

Faculty of Computer Science > Departemen of Data Science

Depositing User:

Divanda Shaffa Ameera

Date Deposited:

27 May 2025 06:38

Last Modified:

27 May 2025 06:38

URI:

https://repository.upnjatim.ac.id/id/eprint/36495

Actions (login required)

View Item