Adapted methods for clustering large datasets of mixed units