In tis paper we propose a methodology to anonymize multidimensional individual data. The goal is to be able to protect data against the reidentification risk. The proposed solution is based on a coclustering method. The coclustering is used to build an aggregated representation of the data, then the model is used to draw synthetic individual data. We show that these synthetic data preserve sufficient information to be used in place of the real data. Finally the protection against the reidentification risk is evaluated.