Chapter 7. Mixture Models
One common approach to model building is about combining or mixing simpler models to obtain more complex ones. In statistic, this type of models are generically known as mixture models. Mixture models are used for different purposes such as directly modeling subpopulations or as a useful trick to handle complicated distributions that cannot be described with simpler distributions. In this chapter we are going to learn how to build them. We are are also going to find that some models from previous chapters were mixture models in disguise, now we are going to uncover them using a mixture-model perspective.
In this chapter, we will learn:
- Finite mixture models,
- Zero-Inflated Poisson distribution
- Zero-Inflated Poisson regression
- Robust logistic regression
- Model-based clustering
- Continuous mixture models
