Model Use Flashcards
Use
What are the four ways to use a model?
Forward modelling, inverse modelling/parameter estimation, scenario modelling (what if), optimisation
Explain what model prediction is
Extrapolating a model to a new x_N+j to predict the future
Explain what inverse modelling is
Back solving to find a value within a lumped parameter
Explain what scenario modelling is
Identifying scenarios and predicting the future for each scenario (e.g do nothing, do more, do less)
Explain what optimisation is
Choose decision parameters to minimise the cost function. e.g what is the maximum ___ while avoiding impacts on ____?
Optimisation is a form of minimisation. What does it mean when there is more than one minima? Is there a way to rank minima?
There can be multiple combinations of the decision parameters (energy example e.g energy efficiency, insulation, HVAC system, lighting) that minimise the objective function. Minima could be ranked by opinions of the client, shareholders, etc, or use engineering judgement.
Explain how model predictions can be used as a learning tool to understand a complex system
Describe how scenario modelling can be used to find the critical parameter that separates two distinct classes of future behaviour
The scenarios can be chosen in a way that only changes one of the parameters of the model. These can then be compared to see which change in parameter is critical - meaning there is the biggest difference when changed.
How are predictions and scenario modelling similar and how are they different?
Both predictions and scenario modelling predict the future. However, predictions use one set of inputs/parameters to find one outcome. Scenario modelling changes inputs/ parameters to create multiple possible outcomes.
How are optimisation and inverse modelling similar and how are they different?
?
Should the intended end use of a model be considered when designing the model itself? i.e from How to Improve
yes but possible bias?
