The main objective of this proposal is to develop a method that provides an interpretable and effective graphical representation of asymmetric proximity matrices (flow data) while preserving information about the intrinsic asymmetry of the relationships. The proposal is based on the transformation of the observed data into a three-way array. The Candecomp/Parafac model is then applied to this transformed structure, allowing a simple analysis of the asymmetries in the data. An artificial example is presented to illustrate the method.