1.4 Which model is best?

All three of these scenarios describe different modeling scenarios. With the saturating and logistic models (Models 2 and 3) we have some limiting behavior the possibility that the the rate of infecion slows. Of the two models, which one is the best one? Here could be some possible criteria we could evaluate:

• Do the model outputs match the data?
• (For timeseries data) are the trends accurately represented?
• Is the model easy to use?
• How will model outputs compare with newly collected measurements?
• Regarding model complexity - how many equations do we have?
• Are the number of model parameters too few or to many?

We will address several of these critera later on in this textbook when we discuss model selection (Section 14). Model selection is one key part of the modeling hypothesis - where we investigate the implications of a particular model analyzed. If we don’t do this, we don’t have an opportunity to test out what is plausible and what is believeable in our models.