PERBANDINGAN KINERJA ALGORITMA TEMPORAL FUSION TRANSFORMER (TFT) DAN LIGHT GRADIENT BOOSTING MACHINE (LIGHTGBM) DALAM PREDIKSI INFLASI DI INDONESIA

AUDYNA RENATA, . (2025) PERBANDINGAN KINERJA ALGORITMA TEMPORAL FUSION TRANSFORMER (TFT) DAN LIGHT GRADIENT BOOSTING MACHINE (LIGHTGBM) DALAM PREDIKSI INFLASI DI INDONESIA. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text final_COVER.pdf Download (1MB)
	Text BAB 1.pdf Download (305kB)
	Text BAB 2.pdf Restricted to Registered users only Download (927kB) | Request a copy
	Text BAB 3.pdf Restricted to Registered users only Download (565kB) | Request a copy
	Text BAB 4.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB 5.pdf Restricted to Registered users only Download (277kB) | Request a copy
	Text Daftar Pustaka.pdf Download (245kB)
	Text Lampiran.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Inflasi merupakan kenaikan harga barang dan jasa dalam periode waktu tertentu. Inflasi menjadi indikator penting yang mencerminkan kestabilan ekonomi suatu negara. Fluktuasi inflasi dapat memengaruhi kebijakan moneter, daya beli masyarakat, serta pengambilan keputusan ekonomi. Penelitian ini bertujuan untuk membandingkan kinerja model Temporal Fusion Transformer (TFT) dan Light Gradient Boosting Machine (LightGBM) dalam memprediksi tingkat inflasi di Indonesia. Data yang digunakan berupa deret waktu bulanan multivariat periodeJanuari 2010 hingga Desember 2024, mencakup variabel inflasi, nilai tukar, suku bunga, ekspor, impor, indeks harga konsumen, harga minyak dunia, dan jumlah uang beredar (M2). Proses penelitian meliputi tahap preprocessing data, pembagian data menjadi data latih dan uji, serta penyesuaian hyperparameter pada kedua model. Evaluasi kinerja dilakukan menggunakan tiga metrik utama, yaitu Mean Absolute Error (MAE), Root Mean Square Error (RMSE), dan Mean Absolute Percentage Error (MAPE). Hasil menunjukkan bahwa LightGBM memiliki kinerja lebih baik dengan MAE = 0,53, RMSE = 0,75, dan MAPE = 15,30%, dibandingkan TFT dengan MAE = 1,16, RMSE = 1,47, dan MAPE = 34,34%. Temuan ini membuktikan bahwa model berbasis boosting lebih efisien untuk dataset ekonomi berskala terbatas, sedangkan model deep learning seperti TFT memerlukan data lebih besar dan konteks temporal lebih kompleks. Dengan demikian, LightGBM dinilai lebih unggul secara empiris dalam memprediksi inflasi di Indonesia. ***** Inflation refers to a sustained increase in the general price level of goods and services over a given period. Inflation is an important indicator of a country's economic stability. Inflation fluctuations can affect monetary policy, public purchasing power, and economic decision-making. This study aims to compare the performance of the Temporal Fusion Transformer (TFT) and Light Gradient Boosting Machine (LightGBM) models in predicting inflation rates in Indonesia. The data used is a multivariate monthly time series for the period January 2010 to December 2024, including the variables of inflation, exchange rates, interest rates, exports, imports, the consumer price index, world oil prices, and the money supply (M2). The research process includes data preprocessing, data division into training and test data, and hyperparameter adjustments for both models. Performance evaluation is conducted using three primary metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE).The results show that LightGBM outperforms TFT with MAE = 0,53, RMSE = 0,75, and MAPE = 15,30%, compared to TFT with MAE = 1,16, RMSE = 1,47, and MAPE = 34,34%. These findings demonstrate that boosting-based models are more efficient for limited-scale economic datasets, while deep learning models like TFT require larger data sets and more complex temporal contexts. Thus, LightGBM is considered empirically superior in predicting inflation in Indonesia.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Irma Permata Sari, S.Pd., M.Eng. ; 2). Ali Idrus, S.Kom., M.Kom.
Subjects:	Teknologi dan Ilmu Terapan > Teknik Komputer
Divisions:	FT > S1 Sistem dan Teknologi Informasi
Depositing User:	Audyna Renata .
Date Deposited:	23 Jan 2026 06:47
Last Modified:	23 Jan 2026 06:47
URI:	http://repository.unj.ac.id/id/eprint/63522

Actions (login required)

View Item