Riswanda, Mohammad Nizar (2025) IMPLEMENTASI MODEL N-BEATS (NEURAL BASIS EXPANSION ANALYSIS FOR TIME SERIES) DENGAN OPTIMASI TREE-STRUCTURED PARZEN ESTIMATOR UNTUK PREDIKSI INFLASI DI JAWA TIMUR. Undergraduate thesis, Universitas Pembangunan Nasional "Veteran" Jawa Timur.
![]() |
Text (Cover)
FIX_cover_repo.pdf Download (1MB) |
![]() |
Text (Bab 1)
FIX_Bab1_repo.pdf Download (249kB) |
![]() |
Text (Bab 2)
FIX_Bab2_repo.pdf Restricted to Repository staff only until 22 July 2028. Download (841kB) |
![]() |
Text (Bab 3)
FIX_Bab3_repo.pdf Restricted to Repository staff only until 22 July 2028. Download (548kB) |
![]() |
Text (Bab 4)
FIX_Bab4_repo.pdf Restricted to Repository staff only until 22 July 2028. Download (5MB) |
![]() |
Text (Bab 5)
FIX_Bab5_repo.pdf Download (221kB) |
![]() |
Text (Daftar Pustaka)
FIX_Dafpus_repo.pdf Download (224kB) |
![]() |
Text (Lampiran)
FIX_lampiran baru_repo.pdf Restricted to Repository staff only until 22 July 2028. Download (4MB) |
Abstract
Inflation is one of the main factors influencing the economic growth of a region. High inflation can lead to economic instability, increased poverty, and decreased purchasing power of the population. Sound economic growth can be achieved if supported by stable inflation, which in turn helps improve public welfare. Therefore, inflation forecasting is essential to assist governments in planning appropriate economic strategies. This study employs the Neural Basis Expansion Analysis for Time Series (N-BEATS) model optimized with the Tree-Structured Parzen Estimator (TPE) to forecast inflation in East Java. The dataset used in this research is univariate inflation data from East Java, covering the period from January 2005 to December 2024, obtained from the official website of BPS (Statistics Indonesia) East Java. N-BEATS is a deep learning model designed for time series forecasting by utilizing a fully connected architecture and stacked blocks, allowing the model to learn from previous errors. The performance of the N-BEATS model heavily depends on key hyperparameters such as the number of neurons, layers, block size, and forecast horizon length. Improper selection of these hyperparameters can negatively impact model performance. To address this issue, TPE optimization is used to enhance model performance by efficiently searching the hyperparameter space based on the model’s prior performance. The results show that the unoptimized N-BEATS model yielded a MAPE of 19.01%, while the N-BEATS model optimized with TPE achieved a lower MAPE of 15.27%. This demonstrates the effectiveness of TPE in improving the performance of N-BEATS for forecasting inflation in East Java.
Item Type: | Thesis (Undergraduate) | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Contributors: |
|
||||||||||||
Subjects: | T Technology > T Technology (General) | ||||||||||||
Divisions: | Faculty of Computer Science > Departemen of Data Science | ||||||||||||
Depositing User: | Mohammad Nizar Riswanda | ||||||||||||
Date Deposited: | 22 Jul 2025 07:27 | ||||||||||||
Last Modified: | 22 Jul 2025 07:27 | ||||||||||||
URI: | https://repository.upnjatim.ac.id/id/eprint/40472 |
Actions (login required)
![]() |
View Item |