On the classification of experimental data modeled via a stochastic leaky integrate and fire model through boundary values