Análisis y predicción de sequías mediante Random Forest para la optimización de la situación energética del país

Abstract

This research paper presents drought analysis and prediction using Random Forest within a drought prediction system. The country's electricity generation depends primarily on 72% from hydroelectric plants, which are vulnerable to climate variability. The study proposes developing a prediction model based on historical data from NASA, thas helps energy management. The drought prediction model was developed using Random Forest, integrating real climate data to calculate the Standardized Precipitation Index (SPI1, SPI3, SPI6). The model was then trained using Random Forest. The results are displayed through a dashboard that incorporates metrics, national SPI, future predictions, and generation data from the country's most important hydroelectric units. The standardized precipitation index (SPI1, SPI3, SPI6) is calculated by the model, which was trained using historical data from NASA. An accuracy of 88.9%, an F1-Score of 87.5%, and a recall of 86.1% were achieved, enabling reliable forecasts to be made three months in advance.