Statistics
- Statistics are really well understood.
- There is a sound foundation for them, and they can be
used for learning.
- Roughly, given past experience (data), use statistical methods
to predict future behaviour.
- There are a lot of different statistical methods, and they
can be used for lots of different problems. Here are some
algorithms and applications.
- Linear Regression can be used for series prediction, and
general prediction.
- K-Nearest Neighbors can be used to cluster items together.
- Bayesian Nets can be used to develop probabalistic models.
- The problem with statistics is that they really depend on
well founded theory.
- As machine learning algorithms are better understood, a statistical
basis is usually discovered.
- So machine learning algorithms are a bit less restricted.
