Model weighting

Model uncertainty can be represented by formulating 2 or more different models to represent alternative hypotheses or viewpoints and then combining the model outputs by assigning weights representing their relative probability or credibility, using either Bayesian and non-Bayesian approaches. 

---

Table 6.16 Rattlesnake Hill tholeiite (Philpotts, 1979) and its unmixes according to Ghiorso-Carmichael model (weight % = total iron as FeO).

A conceptually simple approach that avoids the difficulties of model weighting is scenario analysis or the 1-at-a-time method, where the alternative models are analyzed separately and the resnlts are compared. In the example of the previous section, this might produce a conclusion of the type If model A is true then 0 people will get cancer if model B is trne then 200 people will get cancer. However, this... 

Model weighting different models are combined by assigning weights representing their relative probability, using either Bayesian and non-Bayesian approaches. 

Radiative transfer equation equation of transfer Long plug flow furnace model Source-sink refractory model Well-stirred combustion chamber model Weighted sum of gray gases spectral model... 

Examination of residual error model weighted residuals have unit variance and are uncorrelated so that the shape of distribution about residual = 0 line should be uniform if error model is adequate WRES scale can be viewed as an approximate SD scale (i.e., outside the 3 unit range should be further explored as a potential outlier, see Figure 15.3) Goodness-of-fit and examination of time dependencies can be used to discern tolerance or induction phenomenon Goodness-of-fit and examination of time dependencies residuals should be randomly dispersed about the residual = 0 line... 

Calculation done with a two-component 1 Markovian model. Weight-average molecular weight in Daltons x lO . 

---