Desarrollo e implementación de un algoritmo de predicción de consumo de agua potable y visualización de los datos de consumo por usuario y sectores mediante redes neuronales recurrentes dentro del proyecto CEDÍA-TARPUQ

Abstract

In this work we propose to forecast drinking water consumption using four recurrent neural network architectures of LSTM type, these contain multiple outputs and perform projections to multiple dates, also the best model is chosen through the MAE and RMSE to optimize the forecast of drinking water consumption.