**2. Methodology**

The developed data analytics modules, which aim at improving the operation of the RECs through a better coordination between local generation and consumption, are explained in the present section. More particularly, a local day-ahead wind power forecasting model, able to deal with wind power abnormal data, is presented in Section 2.1. Section 2.2 describes the generator of electricity consumption representative profiles.

#### *2.1. Local Wind Power Forecasting*

This section first describes the different Machine Learning models that are employed for the day-ahead prediction of wind power time series. An original methodology for automatically dealing with abnormal wind power data, which are abundant in the case of localized predictions (as opposed to the case of aggregate predictions made at the regional or national level), during the learning phase of neural network models, thereby improving the forecast performance, is then presented.
