Ismail Gaffar, Arwini Arisandi, Andi Ridwan Makkulawu

Evaluasi Kinerja Model Adaptive Response Rate Exponential Smoothing dalam Memprediksikan Harga Beras

Introduction

Evaluasi kinerja model adaptive response rate exponential smoothing dalam memprediksikan harga beras. Evaluasi model ARRES untuk peramalan harga beras di Makassar. Bandingkan ARRES vs SES, ARRES terbukti lebih akurat (MAPE

Abstract

Penelitian ini bertujuan untuk mengidentifikasi model peramalan terbaik untuk memprediksi harga beras di Kota Makassar dengan membandingkan metode Adaptive Response Rate Exponential Smoothing (ARRES) dengan metode Single Exponential Smoothing (SES). Data harga beras harian untuk beras medium dan premium dari Januari hingga Juni 2025, yang terdiri dari 166 pengamatan, dianalisis menggunakan kedua metode tersebut dengan perangkat lunak RStudio. Analisis data menunjukkan bahwa harga beras premium cenderung lebih tinggi dan lebih stabil dibandingkan harga beras medium. Metode ARRES, yang dilengkapi dengan parameter smoothing adaptif yang merespons kesalahan terbaru, menunjukkan respons yang lebih baik terhadap fluktuasi harga beras dibandingkan metode SES dengan parameter konstan. Evaluasi menggunakan Mean Absolute Percentage Error (MAPE) menunjukkan bahwa ARRES lebih unggul dibandingkan SES, dengan nilai MAPE masing-masing 0,379% dan 0,420% untuk beras medium, serta 0,283% dan 0,317% untuk beras premium. Hasil ini menunjukkan bahwa ARRES memberikan akurasi peramalan yang lebih baik, diklasifikasikan sebagai akurasi sangat tinggi (MAPE<10%), menjadikannya alat yang menjanjikan untuk peramalan harga beras guna mendukung pengambilan keputusan oleh pemerintah dan pemangku kepentingan di Makassar.

Review

This study presents a focused evaluation of the Adaptive Response Rate Exponential Smoothing (ARRES) model's efficacy in predicting rice prices in Makassar, comparing it against the more conventional Single Exponential Smoothing (SES). Utilizing daily price data for medium and premium rice varieties from January to June 2025, the research meticulously analyzes the performance of both methods. A significant finding is the consistent outperformance of ARRES, which, with its adaptive smoothing parameter, demonstrates superior responsiveness to price fluctuations compared to SES. The paper quantitatively substantiates this through Mean Absolute Percentage Error (MAPE) values, indicating ARRES's higher accuracy and classifying its forecasting ability as "very high." The strength of this research lies in its clear methodological comparison and the practical relevance of its findings. By directly contrasting ARRES with SES, the study effectively highlights the advantage of adaptive smoothing in dynamic market conditions. The use of specific data for both medium and premium rice adds valuable nuance, demonstrating that premium rice prices, while higher, exhibit greater stability. The transparent reporting of MAPE values provides concrete evidence of ARRES's improved accuracy, making the results easily interpretable and actionable. The stated implications for government and stakeholders in Makassar underscore the study's potential to inform critical decision-making processes related to food price management and stability. While the study offers valuable insights, a few aspects could be further elaborated or considered for future research. The relatively short data period of six months, though offering granularity, might limit the generalizability of the findings across longer market cycles or periods of greater economic volatility. While ARRES is shown to be superior to SES, the paper does not explore its performance against other advanced forecasting techniques, such as ARIMA models, GARCH, or machine learning algorithms, which might capture additional complexities or external factors influencing rice prices. Future work could also investigate the optimal forecasting horizon for ARRES and explore the integration of exogenous variables to enhance the model's predictive power further.

Full Text

Comments

You need to be logged in to post a comment.