Chapter 6. Model Comparison
"All models are wrong, but some are useful." | ||
| --George Box | ||
We have already discussed the idea that models are wrong in the sense that they are just approximations used in an attempt to understand a problem through data and not a verbatim copy of the real world. While every model is wrong, not every model is equally wrong; some models will be worse than others at describing the same data. In the previous chapters, we focused our attention on the inference problem, that is, how to learn the value of parameters from the data. In this chapter, we are going to focus on a different problem: how to compare two or more models used to explain the same data. As we will learn, this is not a simple problem to solve and at the same time is a central problem in data analysis.
In the present chapter, we will explore the following topics:
- Occam's razor, simplicity and accuracy overfitting, and underfitting
- Regularizing priors
- Information criteria...
