Possible versions of mathematical processing of the experimental data array are considered. It is offered to present them as an approximating function, not passing exactly through all experimental points, but bringing to minimum the square quality criterion J. Parameters of this approximating function (this is often an exponential function) for specific processes are determined by analysis of the results of the experiment on the basis of knowledge of the mechanism of physical processes, being described by the function. An example of treatment of experimental data of weight change of chicken blocks, being defrosted under vacuum, is given.