This project is designed to forecast a plant's electricity consumption. To do this, we have access to a training dataset containing the evolution of electricity consumption over one and a half years, as well as the evolution of 6 factors whose identity is not revealed.
We also have access to a test dataset containing the evolution of the 6 factors over the test period, and we need to predict electricity consumption over this period using the model trained on the training dataset.
We began this project by carrying out an EDA in order to correctly characterize the time series under study. We then trained and tested various models, including SARIMA and LSTM.