For the last decades, Transport Demand and Mobility has been a continuously developing branch in the transport literature. This is reflected in the great amount of papers published on scientific magazines trying to solve the traffic assignment and trip matrix estimation problems. This fact shows the difficulty to elaborate an effective model able to reproduce the real behaviour. These models use several data inputs (prior trip matrix, link counts, etc.), only from a subset of the problem variables, and its size will depend on the available budget. One problem of these models is the high number of possible solutions which is usually solved by finding one where the model results and the real data match up. Nevertheless a model does not have to reproduce only the real data, but also all the variables. Therefore the problem must consist of reproducing the reality in a precise form with minimum cost. The aim of this work consists of proposing new models, which allow us to update the traffic flow predictions from a small subset of real data. To this end, a Gaussian Bayesian Network is used, and so, probability intervals are obtained to get an idea of the associated uncertainties.