In multivariate statistic graphical models are used to visualize the dependence structures of random vectors. We focus on max-linear structural equation models (ML-SEMs) which generate max-linear random vectors. The dependence structures of those vectors can be represented by directed acyclic graphs (DAGs). A ML-SEM is given by a set of weights and random noise variables. The weights are restricted such that they match the edge weights of a DAG. We want to estimate the distribution of a random vector which is generated by a ML-SEM. Therefore, we introduce and investigate an estimator for the weights. Amongst others, we show that the estimator convergences almost sure and that it can be used to learn the DAG behind the model.     «

In multivariate statistic graphical models are used to visualize the dependence structures of random vectors. We focus on max-linear structural equation models (ML-SEMs) which generate max-linear random vectors. The dependence structures of those vectors can be represented by directed acyclic graphs (DAGs). A ML-SEM is given by a set of weights and random noise variables. The weights are restricted such that they match the edge weights of a DAG. We want to estimate the distribution of a rando...     »