Pronóstico del precio diario de bolsa en el mercado eléctrico colombiano usando modelos autorregresivos.

Abstract

The Colombian Wholesale Electricity Market (MEM) exhibits highly dynamic price behavior due to the interaction of operational, hydrological, and demand-related factors, making accurate forecasting a challenging task using conventional methods. This research developed a forecasting model based on autoregressive techniques by implementing a Seasonal Autoregressive Integrated Moving Average with Exogenous Variables (SARIMAX) model to simultaneously capture the temporal dependence of electricity prices and the influence of relevant external variables. The proposed methodology included data acquisition from the Colombian Mining and Energy Information System (SIMEM), exploratory time-series analysis, data preprocessing through outlier treatment, statistical transformations, standardization, lag generation, and feature selection. Subsequently, AR, ARX, SARIMA, and SARIMAX models were implemented and compared using MAE, MSE, and RMSE performance metrics, while the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were employed to identify the most appropriate model structure. The results demonstrated that incorporating exogenous variables, including electricity demand, hydroelectric participation, and reservoir storage levels, significantly improves forecasting performance compared to purely autoregressive approaches. The SARIMAX model achieved the best predictive performance, reducing the RMSE by 18.5% and the MAE by 14.2% relative to the baseline model while maintaining stable behavior during validation. Furthermore, the sensitivity analysis identified electricity demand as the most influential predictor, followed by hydroelectric participation and reservoir storage. Overall, the obtained results confirm that autoregressive models with exogenous variables provide an effective and interpretable approach for forecasting electricity market prices in Colombia, supporting operational planning, market analysis, and decision-making within the National Interconnected System.