Sammendrag
In this thesis a large dataset is investigated with respect to consistency and periodicity with the purpose of being used as basis for prediction. The dataset is then to be used to train a artificial neural network which will be evaluated as a means to predict changes in load for a system.