Statistical Aspects of Kinetic Modeling for Food Science Problems

Abstract

Abstract Statistical techniques to estimate kinetic parameters (rate constants, activation energy, pre‐exponential factor) have been reviewed. Differences between non‐linear and linear regression were indicated. Extended least‐squares was shown to be useful to obtain information about experimental uncertainties of data. Measurement of reactants and products simultaneously (multiresponse) provides the possibility to estimate parameters more accurately than with uniresponse modeling (in which only one reactant or only one product is analyzed). Four examples were used to illustrate: (1) possible bias introduced by linearizing a first‐order equation; (2) use of extended least‐squares; β) advantages of multiresponse modeling; and (4) statistical problems associated with the Arrhenius equation.