Time Series Forecasting using NARX and NARMAX models with shallow and deep neural networks

In this work shallow and deep neural networks are used to develop NARX and NARMAX models for the prediction of two time series with different characteristics. The hypothesis is that the models generated with deep learning techniques outperform shallow techniques. The results show that for problems of medium complexity the proposed hypothesis is fulfilled highlighting in this case the use of convolutional neural network (CNN) On the other hand for problems of low complexity the hypothesis is not fulfilled so in in these cases the use of Extreme Learning Machine (ELM) is recommended.