Symbolic Regression, Bifurcations, and Logistic Models Applied to Commodity Volatility: A Case Study in the Peruvian Electrode Industry

• Autor: Cáceres Linares, Luis César
• Formato: artículo
• Fecha de Publicación: 2025
• Institución: Universidad Nacional Autónoma de Tayacaja Daniel Hernández Morillo
• Repositorio: Revista de investigación científica y tecnológica Llamkasun
• Lenguaje: español
• OAI Identifier: oai:ojs2.llamkasun.unat.edu.pe:article/140
• Enlace del recurso: https://llamkasun.unat.edu.pe/index.php/revista/article/view/140
• Nivel de acceso: acceso abierto
• Materia: Dinámica caótica; Regresión simbólica; Commodities; Índice Dow Jones; Indice LME; Cadena de suministro; Chaotic dynamics; Symbolic regression; Dow Jones index; LME index; Supply chain

This research proposes a methodological framework based on chaotic dynamics and nonlinear equations to analyze the relationship between international stock market indices (Dow Jones and LME) and the price behavior of imported commodities essential for electrode manufacturing in Peru. The volatility of these inputs directly impacts the profitability and sustainability of the national welding industry, especially under highly uncertain global conditions. ARIMA models (Box-Jenkins), symbolic regression (SR), and Verhulst logistic equations were applied to model time series of commodity prices from 2018 to 2023. Additionally, bifurcation analysis and the Feigenbaum constant were used to detect chaotic transitions. Results show that nonlinear models outperform traditional linear approaches, with lower Root Mean Square Error (RMSE) in predictive performance. Empirical validation confirmed that erratic market behavior can be anticipated through dynamic attractors. It is concluded that integrating advanced mathematical tools enhances supply chain management by providing a predictive system that reduces uncertainty in the procurement of raw materials. This approach supports better strategic decision-making in the electrode manufacturing industry and may be applicable to other industrial sectors highly sensitive to commodity price fluctuations.