The non-linear Lev-Mar fitting routines minimize the Chi-Square distance between the data and the model. The Chi-Square distance is just the Least-Squares distance weighted by the inverse of the standard-deviation at a each point. Points that have small standard deviations are presumed to be "better" and are given more weight. The online help for the Nonlinear Lev-Mar Fit.vi has all the relevant definitions and formulae.
In your example you would find the mean and standard deviation of the Y values for a given X. Do this for all X values. You will then have your original array of X values, an array of means (Y values), and an array of standard deviations. These are the inputs to the Lev-Mar VIs, namely "Nonlinear Lev-Mar fit.vi" and Levenberg-Marquardt.vi".
