Beyasian changepoint detection notation

30 Views Asked by At

I'm reading this article and I'm having some problems in understanting basics or notation.

Article is about detecting changepoint with bayesian approach. So we have vector with data $Y = \{y_1, y_2,\ldots, y_n\}$ and if there are $k$ changes in moments $t_1,\ldots, t_k$ we split $Y$ into $(y_1,\ldots,y_{t_1-1}), (y_{t_1},\ldots,y_{t_2-1}),\ldots, (y_{t_k},...y_n)$. And after that my problem begins. In the article we read

In a Bayesian formulation the joint posterior distribution for the latent changepoint indicator vector $z$ and segment parameters $θ = \{θ_1,\ldots,θ_{k+1}\}$ can be written as a product of the full segment likelihood $(2.2)$ and the priors for $z$ and $θ$, (...)

where $\theta$ is a parametr from likelihood estimation.

I'm not a native english speaker so I don't really understand that sentence and so on the next equation

equation from article

that $\pi$ function came from nothing (for me) and what does "$\alpha$" mean?

Greetings