In this work, we present recent advances on the creation of data-driven models to address the needs of transportation planners during the conception, understanding and maintenance of transport networks.More specifically, we present data-driven models to understand and analyse mobility and to simulate and predict the impact of changes in existing networks.This new modelling approach leverages the massive amount of data collected in the field from daily users transactions and sensors outputs, and proposes to use in a more extensive way machine learning techniques that have emerged over the last decade.First, we present how this new transportation modelling approach differs from traditional practices.We then illustrate this approach through some specific use cases where it has been applied and present the preliminary results we have obtained.We finally end up with a discussion highlighting both the main advantages and the high potential of adopting such an approach in the transportation planning domain and also the main obstacles to be overcome before a large adoption can happen.