I have numerous one dimensional vectors, $V_1,...,V_i$. Each vector is of variable size composed of natural numbers from different unknown distributions. I'd like to find a way to group/cluster values in each vector. For each value in $V_i$, I would like to assign a group label to each value or label a value as belonging to no group. Can anyone suggest an approach to group these data sets? I've tries various unsupervised clustering approaches, which all work to some degree, but the variable densities and distributions of each data vector makes the optimal parameters for one vector poorly suited to another vector.