Aplicación de Modelos de Machine Learning para la Predicción de Generación Energética en Perú usando datos del COES

[christian_fonseca]

Abstract

The present study evaluates advanced methods to predict energy generation in Peru using data from the Committee of Economic Operation of the Interconect National System (COES). They compare traditional approaches such as ARIMA models with deep learning techniques, including deep neural networks (DNN) and recurrent neural networks (RNN). Also evaluated are machine learning models such as linear regression, support vector machines (SVM), Random Forest and XGBoost. The preparation of the data included filtration, grouping, correlation analysis and backtesting tests. The results indicated that XGBoost offers the best precision with an MAE of 78.76, followed by Random Forest with an MAE of 162.35. The linear regression and SVM showed an inferior performance with MAEs of 190.52 and 291.25, respectively. They conclude that deep learning models and the increase in computational capacity have revolutionized the prediction of energy generation, recommending the adoption of advanced technologies and the inclusion of more auxiliary factors.

INTEGRANTES (Grupo 7 - Seccion A):
- Mamani Cayo, Eduard
- Parado Sosa, Daniel
- Rosales Fierro, Jesus
- Fonseca Rodriguez, Christian

Link del Codigo: https://colab.research.google.com/drive ... sp=sharing