This is not a nonlinear problem in the unknown coefficients. Use regress, since you have the stats toolbox. You could even use backslash, but regress gives you more information.
help slash
help regress
By the way, lnz is not how you compute the natural log. Use log.
