Skripsi

PERBANDINGAN KINERJA LSTM, RANDOM FOREST, DAN SVR BERBASIS KNOWLEDGE DISCOVERY UNTUK PREDIKSI HARGA BERAS SUMATERA SELATAN

---

Rice is a primary staple food in Indonesia, particularly in South Sumatra Province. In February 2024, BBC News Indonesia reported that the price of premium rice surged to Rp18,000 per kilogram, marking the highest price in the country’s history. To anticipate and predict similar spikes in the future, this study applies a Knowledge Discovery approach and compares three machine learning models: LSTM, Random Forest, and SVR. The approach follows the stages of data selection, cleaning, transformation, modeling, and evaluation to uncover hidden patterns in historical data. The dataset, obtained from the official PIHPS Nasional website, consists of 1,412 daily rice price records from January 2020 to May 2025. Model performance was evaluated using MAPE, MAE, and RMSE metrics. The findings indicate that the SVR model outperformed LSTM and Random Forest, delivering the most accurate results. For the Super Quality II rice category, SVR achieved a MAPE of 0.00 percent, MAE of 40.93, and RMSE of 52.54. SVR also consistently produced the lowest prediction errors in other categories, such as Low Quality I (MAE 59.39) and Medium Quality I (MAE 38.92). This research is expected to serve as a foundation for developing machine learning–based food price monitoring systems to support more responsive policies and maintain rice price stability in the future. Keywords: Rice Price; Long Short Term Memory; Random Forest; Support Vector Regression; South Sumatra

---

Availability

Detail Information

Series Title

-

Call Number

T1895362025

Publisher

Indralaya : Prodi Sistem Informasi, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xvi, 96 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

005.130 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Informasi
Algoritma, Pembelajaran--Pembelajaran Komputer

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment