Multidimensional data warehouses represent an important artefact used by organizations to facilitate decision-making. However, they can be inefficient to cope with the growing need to execute complex queries. In addition, they experience difficulties in managing sparse matrices and are not efficient in managing schemas’ evolution. To deal with these problems, we propose to transform these multidimensional models into NoSQL graph models. Our approach ensures the automatic mapping of multidimensional schemas into property graph models. It consists of a set of mapping rules defined between two meta-models, namely multidimensional and property graph meta models. Finally, we illustrate the application of our approach to a case study and discuss its validation.