Machine Learning-Based Early Warning Indicator for Sawn Timber Price: A Comparative Model Analysis with a Case Study of Korea

View at publisher Summarise with VIRAT
Authors: Jae Hyun Rhee, Chang-Deuk Eom, Byeong-Il Ahn
Cited by: 0
References: 43

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Abstract This study developed an early warning indicator (EWI) for sawn timber prices in South Korea by applying machine learning (ML) models. Specifically, we compared the predictive performance of artificial neural network (ANN), support vector machine (SVM), and random forest (RF) models while employing Bayesian optimization to fine-tune hyperparameters and improve both accuracy and computational efficiency. Using monthly timber market data, the empirical analysis revealed that ANN consistently achieves the highest predictive accuracy across both in-sample and out-of-sample tests. However, when applying crisis thresholds of 1.65 and 1.96, the ANN model tended to yield overcautious signals. In contrast, SVM and RF performed adequately within the training sample but showed reduced robustness and stability in out-of-sample forecasting, particularly under conditions of data scarcity. These findings underscore the practical utility of ANN as a reliable tool for early detection of timber price shocks, thereby offering policymakers a proactive instrument with which to mitigate risks in a volatile market. Beyond empirical contributions, this research enriches the methodological framework for constructing EWIs in the timber sector and demonstrates the potential of data-driven approaches for enhancing market surveillance and risk management in forest policy and economics.

Fields of Study & Subjects

Forecasting Techniques and Applications Stock Market Forecasting Methods Economics of Agriculture and Food Markets

Publication details

Journal / source
Forest Products Journal
Volume
76
Issue
2
Pages
125–135
Publication date
1 March 2026
Language
EN