Agriculture Reference
In-Depth Information
-
Level of precision of outputs
+
Feed
intake
Crude
fat
Fatty
acids
AME
Glucose
Acetyl-coA
AT P
Amino
acid
gain
Protein
gain
Nutrient
oxidation
CO 2
Live weight
LPRm
-
Number of data requiered for calibration
+
Fig. 8.1. Schematic representation of the relation between the level of precision of model outputs and the
number of data required for model calibration.
the growth potential of every genotype and
environment these authors proposed to char-
acterize growth profiles on the basis of field ob-
servations. Growth profiles may therefore be
defined for groups of birds of the same geno-
type and gender and grown under similar
environmental conditions. The adjustment
of the inputs according to the observed per-
formance therefore enables the  user to de-
scribe different growth profiles with no need
to define the environmental conditions. This
approach assists in overcoming the limita-
tions of field data availability to develop a
portfolio of growth profiles that can be ap-
plied in the turkey industry.
density), which may result in increased
model complexity (i.e. number of param-
eters and variables). However, by increasing
the level of model complexity there is a risk
of defining a number of parameters that do
not improve model precision. To limit this
risk, any parameter or variable added to the
model should be justified in the sense that it
should demonstrate the gain in the expected
level of precision or accuracy. Thus, model
development strategy should start from a
simple model, going towards a step-by-step
increasing level of complexity. A continu-
ous improvement cycle could therefore be
considered in model development strategy
by: (i) developing a first simple version
of  the model; (ii) validating this version;
(iii) improving the model accuracy by add-
ing other parameters; and (iv) re-evaluating
model predictions. Also, this strategy helps
minimize the risk of failure while ensuring
that users understand the basics of the
model before progressing to more compli-
cated simulations.
As an alternative to defining genetic po-
tential and environmental conditions as pro-
posed in other models (EFG Software, 1995;
Ferguson, 2006), Rivera-Torres et al . (2011c)
confounded the effect of environment with
genetic potential to facilitate model application
on commercial farms. Rather than defining
Model interface
Defining the user interface is an important
process that needs to be carried out together
with users and decision makers. Models may
either be developed directly using model-
ling software that illustrate the flows and
the compartments (e.g. Rivera-Torres et al .,
2011c) or programmed as independent soft-
ware with a customized interface (e.g. EFG
Software, 1995). Modelling software may be
preferred for research purposes, while an
independent software tool is the ideal solu-
tion for commercial application. To date,
 
Search WWH ::




Custom Search