Week One Flashcards
Models are abstractions because…
They purposely omit data (the truth is so complex its unknowable), lets you know what the modeler thinks is important, and are used to make completely tractable.
A good model’s distance is…
shorter, because it is closer to the truth. Less distance = closer to truth.
What is the IT approach?
A theoretical approach that uses model selection uncertainty to improve decision making.
Why is looking at just one model bad?
It cuts out other important data that other models could be telling us.
What is the principle of parsimony?
The balance between simplicity and complexity.
Bias is…
too few parameters.
Bias can cause…
the model to be overly influenced by the too few parameters chosen.
Variance is…
too many parameters.
Variance can cause…
the model to start fitting the “noise”, rather than the “signal” of data.
What is an appropriate model?
An approximation of the truth developed to reflect the data. They are often nested, subjective, make biological “sense”, and are supported by good data and sampling.
“a priori” is to…
come up with the model before you even fit or analyze the models.
Define AIC.
The measure of the relative distance between the model and truth. The smaller, the better. Has no actual units.
Define ΔAIC.
The difference between the given model’s AIC measurement and the best model’s AIC measurement.
What is the AIC of the “best model”?
The one with the lowest AIC. Ideally, 0.
Define Wᵢ
The strength of evidence for being the optimal model. All models’ Wᵢ must add to 1.
