Models choice

However, regression coefficients cannot be used to determine the adequacy of a model choice, as discussed previously. 

An operational description of model applications and user influence on model choice. 

Transparent being well-documented, including the reasoning for model choices. 

In this case, the residual will be correlated with the weighted sum of the three partial pressures. Although a trend may still be detectable, we will not generally be able to determine with which of the variables the residual is correlated and thus will not know how to correct the model a careful initial model choice circumvents this problem. 

Before the viscosity can be calculated from capillary data, as mentioned above, the apparent shear rate, 7 , must be corrected for the effect of the pseudoplastic nature of the polymer on the velocity profile. The calculation can be made only after a model has been adopted that relates shear stress and shear rate for this concept of a pseudoplastic shear-thinning material. The model choice is a philosophical question [11] after rheologlsts tried numerous models, there are in general two simple models that have withstood substantial testing when the predictions are compared with experimental data [1]. The first Is ... 

Techniques for parameter estimation vary considerably. If consistent values for model parameters cannot be obtained, the investigators may decide that the model is itself unreliable and should be changed. Thus, model choice and parameter estimation are interactive. A number of workers have discussed generalised procedures [16—18]. Yeh [19] developed numerical algorithms and showed that multiple linear regression could be used successfully if the reaction scheme consisted of steps such as those shown in eqn. (41) or... 

Figure 17.4 enlarges the breakthrough parts at X<0.1 that is below 10% of the feed. If we choose a given percentage of the feed, we can see that the models foresee different service times. Thus the model choice is of primary importance for calculating service time of cartridges. 

In the development of our 3-D QSAR models, choice of analogue conformation played an important role in providing a realistic pharmacophore model. The... 

Wheeler JR, Grist EPM, Leung KMY, Morritt D, Crane M. 2002a. Species sensitivity distributions data and model choice. Mar Poll Bull 45 192-202. 

One possible model choice for p(k) that is of widespread use in statistical applications, because of its simplicity and flexibility, is the two-parameter gamma distribution 13... 

In the MATLAB Statistics Toolbox, there are two functions for generating exact D-optimal designs, cordexch and rowexch. Both procedures are equivalent from the user s point of view. To use them, one must specify the number of variables, the number of the experiments, and the type of the desired regression model. Four different model choices are provided ... 

Whether the isolated layer model or the Donnan-like model is more appropriate depends on whether the clay particles are considered to be mainly isolated layers or mainly floes. A montmorillonite dispersion, for example, tends to be formed by isolated layers at high pH and low electrolyte concentrations, especially when the electrolyte cation is Li+ or Na+, but it is usually aggregated at low pH and high electrolyte concentrations. Therefore, in a simple titration experiment between pH 3 and 10 at low electrolyte concentration, the aggregation state can change from Hoc to isolated layer, complicating the choice of model. However, although they are conceptually different, both models should perform similarly with appropriate parameters, and the final model choice is mainly a matter of taste or mathematical convenience. 

Patterns of exposure can be described using models that combine abiotic ecosystem attributes, stressor properties, and ecological component characteristics. Model selection is based on the model s suitability for the ecosystem or component of interest, the availability of the requisite data, and the study objectives. Model choices range from simple, screening-level procedures that require a minimum of data to more sophisticated methods that describe processes in more detail but require a considerable amount of data. 

Broens, Bargeman and Smolders( ) reported on the use of nitrogen sorption/desorption method for studying pore volume distributions in ultrafiltration membranes. The pore volume distributions were calculated for a cylindrical capillary model. More recent results from the same laboratory are published in this volume ( ). In our view, applicability of cylindrical pore models for asymmetric membranes should be verified, rather than assumed. This can be done, for example, by analysis of both branches of the sorption isotherm. For a reasonable model choice, the two pore volume distributions should be in substantial agreement. 

B. P. Carlin and S. Chib, Bayesian model choice via Markov chain Monte Carlo methods. / Roy Stat Soc Br 57 473 84 (1995). 

We first discuss the influence of structural model choice and the need of its prespecification in BE assessment. On a rough scale, the AUC (and thus the ratio of test/reference) estimate should be relatively robust, because a model tends to represent the average concentration reasonably well. However, for C ax the opposite should hold. For an illustration, assume that the data arose from a two-compartment model with first-order absorption. Fitting a one-compartment model to the data would underestimate Cmax and thus likely obscure any potential difference of Cmax in the test and reference formulations. Therefore, using a less sophisticated model is likely to bias BE assessment (of Cmax) toward equivalence. 

Base Model Choice. The choice was a steady-state one-compartment model with first-order absorption or a steady-state oral two-compartment model with first-order absorption. The disposition parameters were to be expressed in volume and clearance. Intersubject variability and residual error were also to be assessed. The best-fit model, using the software NONMEM, was to be the final base model. The criteria for accepting the NONMEM base model included (a) improved fitting of the diagnostic scatterplots (observed vs. predicted concentration, residual/weighted residual vs. predicted concentration... 

Model choice is an important sonrce of nncertainty for the purpose of quantitative risk assessment. Changing the underlying modeling assumptions can have a dramatic effect on the estimated benchmark dose. The committee snggests that the K-power (K > 1) model results be used. 

---