Make thorough use of your model, running it over again and always checking
your expectations against its results. Change the initial conditions and try running
the model to its extremes. At some point you will want to perform a formal
sensitivity analysis. Later, we will discuss STELLA’s excellent sensitivity analysis
procedures and other features to complement your modeling skills.
1.6 Principles of Modeling
Though our title of this section may seem somewhat ostentatious, we surely have
learned something general about the modeling process after many years of trying.
So here is our set of ten steps for the modeling process. We expect you to come back
to this list once in a while as you proceed in your modeling efforts, and to challenge
and refine these principles. A good set of principles should be useful to the novice
and aid in speeding the process of learning to become an effective modeler.
1. Define the problem and the goals of the model. Set the questions you want the
model to answer. The power of a good set of specific questions is hard to
overstate. Good questions focus the mind on some aspect of the entire system in
which your subsystem of interest is embedded. Appropriate generalization will
come with time. Spend a lot of time defining the question(s) to be answered by
your model.
2. Select the state variables, those variables that are to be the indicators of the
status of the system through time. Designate the condition for non-negativity of
the state variables, as appropriate. Some state variables are conserved, some are
not. Identify those in your model. Keep the number of state variables as small
as possible. Purposely avoid complexity in the beginning. Record the units of
the state variables either in the “Units” editor available in the side-docked panel
or by adding them in a comment within braces—{. . ..}—in your specification
of the state variable.
3. Designate the control variables, those flow controls that will change the state
variables. Note which state variables are donors and which are recipients with
regard to each of the control variables. Note whether lagged effects should be
included either in the controls or in the variables that compose the controls. Be
sure to set these flows as biflows if appropriate. Also, note the units of the
control variables as you have done for state variables, and ensure that they
match with their uses in other parts of the model. Are there any useful analogies
to apply here? Keep it simple at the start. Try to capture only the essential
features. Put in one type of control as a representative of a class of similar
controls. Add the others as needed, in step 10.
4. Select the parameters for converters. Note the units of these converters and
check their consistency with other parts of the model. Ask yourself: Of what are
these converters a function? Do you expect some of these variables to be lagged
or delayed functions of some of the other variables? Only begrudgingly expand
your model.
22
1 Modeling Dynamic Biological Systems
your expectations against its results. Change the initial conditions and try running
the model to its extremes. At some point you will want to perform a formal
sensitivity analysis. Later, we will discuss STELLA’s excellent sensitivity analysis
procedures and other features to complement your modeling skills.
1.6 Principles of Modeling
Though our title of this section may seem somewhat ostentatious, we surely have
learned something general about the modeling process after many years of trying.
So here is our set of ten steps for the modeling process. We expect you to come back
to this list once in a while as you proceed in your modeling efforts, and to challenge
and refine these principles. A good set of principles should be useful to the novice
and aid in speeding the process of learning to become an effective modeler.
1. Define the problem and the goals of the model. Set the questions you want the
model to answer. The power of a good set of specific questions is hard to
overstate. Good questions focus the mind on some aspect of the entire system in
which your subsystem of interest is embedded. Appropriate generalization will
come with time. Spend a lot of time defining the question(s) to be answered by
your model.
2. Select the state variables, those variables that are to be the indicators of the
status of the system through time. Designate the condition for non-negativity of
the state variables, as appropriate. Some state variables are conserved, some are
not. Identify those in your model. Keep the number of state variables as small
as possible. Purposely avoid complexity in the beginning. Record the units of
the state variables either in the “Units” editor available in the side-docked panel
or by adding them in a comment within braces—{. . ..}—in your specification
of the state variable.
3. Designate the control variables, those flow controls that will change the state
variables. Note which state variables are donors and which are recipients with
regard to each of the control variables. Note whether lagged effects should be
included either in the controls or in the variables that compose the controls. Be
sure to set these flows as biflows if appropriate. Also, note the units of the
control variables as you have done for state variables, and ensure that they
match with their uses in other parts of the model. Are there any useful analogies
to apply here? Keep it simple at the start. Try to capture only the essential
features. Put in one type of control as a representative of a class of similar
controls. Add the others as needed, in step 10.
4. Select the parameters for converters. Note the units of these converters and
check their consistency with other parts of the model. Ask yourself: Of what are
these converters a function? Do you expect some of these variables to be lagged
or delayed functions of some of the other variables? Only begrudgingly expand
your model.
22
1 Modeling Dynamic Biological Systems