03 June, 2010

All models are wrong

My research deals a lot with different analytic models. Here are two quotes about that, a long one:

"Unfortunately, models are rarely exact. The semblance of sophistication inherent in the model and used to develop parameter estimates frequently masks their deficiencies. Models are only approximate, and their predictions when the parameter estimates are based on analysis of plant performance must be considered as approximate. Validation of the model and the parameter estimates using other operating conditions will reduce the likelihood that the conclusions have significant error."

Perry's Chemical Engineers' Handbook


But there is no need in long quotes since we have a short one that captures everything:

"Essentially all models are wrong, but some are useful."

George Box


Take care,

Misha

No comments: