PERAMALAN INDEKS HARGA KONSUMEN KOTA SURABAYA MENGGUNAKAN METODE SVR DAN XGBOOST

Riswanda, Mohammad Nizar and Saputra, Yayang Dimas (2025) PERAMALAN INDEKS HARGA KONSUMEN KOTA SURABAYA MENGGUNAKAN METODE SVR DAN XGBOOST. Project Report (Praktek Kerja Lapang). Universitas Pembangunan Nasional "Veteran" Jawa Timur, Surabaya. (Unpublished)

	Text (Cover) PKL_COVER_REPO.pdf Download (1MB)
	Text (Bab I) PKL_BAB1_REPO.pdf Download (16kB)
	Text (Bab II) BAB 2.pdf Restricted to Repository staff only until 22 July 2028. Download (101kB)
	Text (Bab III) BAB 3.pdf Restricted to Repository staff only until 22 July 2028. Download (544kB)
	Text (Bab IV) BAB 4.pdf Restricted to Repository staff only until 22 July 2028. Download (7kB)
	Text (Daftar Pustaka) DAFPUS.pdf Download (7kB)
	Text (Lampiran) LAMPIRAN.pdf Restricted to Repository staff only until 22 July 2028. Download (3MB)

Abstract

This report discusses research aimed at forecasting the Consumer Price Index (CPI) of Surabaya City using two machine learning methods: Support Vector Regression (SVR) and Extreme Gradient Boosting (XGBoost). CPI is a crucial economic indicator that reflects inflation and regional economic stability. Given Surabaya’s role as a major metropolitan area in Indonesia, its CPI exhibits complex dynamics that require robust modeling for accurate forecasting. The study utilizes historical CPI data from the Surabaya City Statistics Agency (BPS), spanning the years 1998 to 2024. This dataset is analyzed to develop predictive models based on machine learning techniques. The results indicate that SVR outperforms XGBoost in forecasting accuracy, as shown by evaluation metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE). Specifically, SVR achieved lower error values across all metrics: for MSE, XGBoost recorded 0.5935 while SVR posted only 0.0462; for MAE, XGBoost reached 0.4123 compared to SVR's 0.1992; and for MAPE, the results were relatively close, with XGBoost at 0.0037 and SVR slightly better at 0.0019. These findings confirm that SVR offers higher predictive accuracy than XGBoost in modeling Surabaya’s CPI. This research underscores the potential of machine learning in economic forecasting and is expected to contribute to more informed and effective economic policy-making.

Item Type:

Monograph (Project Report (Praktek Kerja Lapang))

Contributors: