Forecasting the Short-Term Energy Consumption Using Random Forests and Gradient Boosting

TL;DR

This study compares Random Forests and Gradient Boosting for short-term energy consumption forecasting, demonstrating that combining them with a weighted ensemble yields more accurate predictions than individual models.

Contribution

It introduces a weighted ensemble approach to improve energy consumption forecasting accuracy over single algorithms.

Findings

Weighted ensemble outperforms individual models

Random Forests and Gradient Boosting are effective for energy forecasting

Ensemble method improves prediction accuracy

Abstract

This paper analyzes comparatively the performance of Random Forests and Gradient Boosting algorithms in the field of forecasting the energy consumption based on historical data. The two algorithms are applied in order to forecast the energy consumption individually, and then combined together by using a Weighted Average Ensemble Method. The comparison among the achieved experimental results proves that the Weighted Average Ensemble Method provides more accurate results than each of the two algorithms applied alone.