Space weather processes, in general, are non-linear and time-varying. In such cases ‘data driven models’ such as Neural Network, Fuzzy Logic and Genetic Algorithm based models were proved promising to be used in parallel with the mathematical models based on first physical principles. In particular, with the recent developments in ‘big data’ systems, one of the urgent issues is the development of new signal processing techniques to extract manageable, representative data out of the ‘relevant big data’ to be employed in ‘training’, ‘testing’ and validation phases of model construction. Since 1990, under the EU Frame Work Program Actions, we have developed such models for nowcasting, forecasting, warning and also for filling the data gaps on space weather cases including prediction of orbital spacecraft parameters. In particular, some typical, illustrative examples include the forecasting of the ionospheric critical frequencies foF2, during disturbed conditions, such as solar storms and extreme events; GPS total electon content(TEC); solar flare index during solar maximum and the construction of solar EUV flux variations. The associated input data organisation and the typical errors which have been within the acceptable operational expectations are summarised in terms of absolute values, percent and RMS. The aim of the paper is to show that the data driven approaches are promising for the forecasting of space weather.