# Bayesian Linear Regression is so hard to understand?

I'm learning Bayesian Linear Regression from a book, the linear model is $$p(w|x,\phi,\sigma^2)=Norm_w[\phi^Tx,\sigma^2]$$, as put in the book, we use Bayes approach to do the parameters estimation.

Here comes the problem: I thought it is pretty clear that we should introduce a conjugate prior for the parameters $\phi$ and $\sigma^2$, which should have a normal-scaled inverse gamma distribution, right? But the book first assume that $\sigma^2$ is known, and introduce a prior distribution for $\phi$ alone, which is a 0-mean Gaussian, and does the estimation for $\phi$. After all that, it assumes $\sigma^2$ is not known, and estimates it.

Why should we separate them?

UPDATE

The book is Machine Learning: A Probabilistic Perspective, p. 232, Section "Baysian Linear Regression."

I just found this article which also assumes one is known and later assumes it is unknown. Bayesian Linear Regression

Alternatively, the purpose could be to compare how inference on the parameters is influenced by the choice of priors on $\sigma^2$, compared to a scenario where $\sigma^2$ is assumed known.
• As said in the book, it assumes $\sigma^2$ is known indeed for simplicity, but the prior will be different when assume $\sigma^2$ is know or not, so I don't see the point why should it make such assumption. – avocado Dec 4 '13 at 13:54