Prediction of temporal series with recurrent neural networks

Abstract

The objective of this project is to program a set of recurrent neural networks (RNN) that are useful to predict temporal series and apply this to different real or simulated cases where we want to predict the future values of a magnitude with previous data. This project is divided into different parts that start with finding a real or simulated situation where we can apply its technique. Then we have to find or generate some data to prove that the method works in this situation with whatever adaptations are necessary and finally we will test the RNN output and we will assess the results and evaluate the error. This will allow us to validate if it is good enough or it is not sufficient and, therefore, we need to upgrade the code