Forecasting the Behavior of Electric Power Supply at Yucatan, Mexico, Using a Recurrent Neural Network

The forecast of electric power generation and supply with respect to an expected demand is a matter of national strategy for countries around the world as well as of vital importance for the assurance and viability of current societies. In the state of Yucatan, Mexico, power generation authorities often experience overproduction due to estimations that are done based on historical data in an statistical manner. In this work, we propose the implementation of a long short-term memory recurrent neural network to predict the consumption of electrical power in the aforementioned state. The main outcome shows that this approach implies a reduction in the error of the estimations of 39.53% provided by the neural network forecast with respect to previous estimations by local power generation experts.