|
Multivariate prediction uncertainty |
|
|
|
|
Written by Klaas Faaber, Chemometry Consultancy
|
Calibration models are at best (local) approximations, and so are the predictions based on such a model. This is a fact that does not depend on the complexity of the input data, i.e. it should not matter whether the input data are univariate or multivariate. Model–based results such as predictions must therefore be reported together with an estimate of their uncertainty.
In a rather provoking editorial [1], De Bièvre expressed this requirement as  follows: “So, a result without reliability (uncertainty) statement cannot be published or communicated because it is not (yet) aresult. I am appealing to my colleagues of all analytical journals not to accept papers anymore which do not respect this simple logic.”
|
Tutorial