Nowadays, data are generated and collected in many domains from various sources. In most of the cases, they are handled as common data where some simple calculations are used to analyse them as measuring the average, the maximum, the deviation, etc. For instance, the average number of children in European families is 1.8 children. This kind of assessment is far away from reality: the number of children should be an integer number. For this reason, exemplars have a finer meaning since its aim, in this case, is to look of an exemplar of a common family in Europe which has 2 children (the most representative family). The aim of this paper is to propose a methodology able to extract representative exemplars from a dataset. This methodology has been experimented with dataset extracted from experimentations of connected vehicle traces. This data analysis has shown some interesting features: the vehicle connectivity guarantees that messages are not lost.