An Hourly Electricity Demand Forecaster for the Mexican National Interconnected System based on Regional Statistical Features

Abstract

This document presents a forecaster based on statistical features to implement the electricity load forecasting for the Mexican National Interconnected System. A particularity of this work is that the data used in the forecaster implementation are those produced by the National Energy Control Center. The characteristics extracted from the data helps to build a more robust forecaster. The results obtained shows good accuracy with the set of features extracted from each time series of the different regions. Therefore, this work can be extended in order to achieve better performance by extracting more features for the data as well as including features outside the time series such as bank holidays, bank days, and so on. It has to be remarked the performance of the forecaster could have been affected by the COVID 19 pandemic as there were many strange behaviors in the energy demand. It is expected that once everything return to normal, the efficiency of this forecaster will improve just because of the data regularity.