Day One part 2

The model building process

 * A clear objective
 * A set of assumptions
 * Usually implied by objective and state of knowledge/data
 * Includes explicit and excluded components and relationships
 * Formal conceptual model
 * Implementation
 * Data requirements and parameterization
 * Output and analyses

Spatio-Temporal Models: A Dozen Deadly Pitfalls
Pitfall #1
 * Failure to define an achievable goal
 * Know why you are modelling
 * A model should never be pressed to do, nor criticized for failing to do, that for which it was never intended (Smith)

Pitfall #2
 * Mistakes you make in the first 10 minutes will return to haunt you for the remainder of the project (C. Walters)
 * work out a solid, semiformal conceptual model before implementation
 * use prototypes to try out ideas and test assumptions

Pitfall #3
 * Incomplete mix of essential skills
 * Project leadership
 * Conceptual modeling
 * Model implementation
 * System knowledge

Pitfall #4
 * Inadequate levels of user participation
 * Who is the model being built for?

Pitfall #5
 * Garbage in – garbage out
 * a model cannot be any better than the information that goes into it

Pitfall #6
 * KISS: Keep it simple, stupid
 * Small models are beautiful (Ludwig)
 * Don’t create a model you can’t understand (Lertzman)
 * Do not build a complicated model when a simple one will suffice (Smith)

Pitfall #7 “There is a tendency to spend a great deal of effort modeling in unnecessary detail those portions of the system that are well understood, while glossing over poorly defined portions that may be more important” (CACI)
 * Inappropriate level of detail and system bounding
 * Scale and resolution a critical

Pitfall #8
 * Beware of molding the problem to fit the technique (Smith)
 * modeling language/platform should fit the model requirement

Pitfall #9
 * Hidden assumptions
 * Documentation is the road to salvation
 * Modelling can be viewed as exploring the consequences of our assumpions
 * This is only useful if assumptions are explicit

Pitfall #10
 * Using an unverified model
 * Verification, validation and sensitivity analysis are not optional
 * Attempt to reject your model, not support it (Lertzman)

Pitfall #11
 * A model should never be taken too literally – models cannot replace decision-makers (Smith)
 * A model does not replace critical thinking
 * never trust the results of a model unless you can explain them
 * Beware or equifinality
 * …where a given end state can be reached by many potential means

Pitfall #12
 * Poor communication
 * Assumptions
 * Results