$$\overbrace{p(\theta|d, \mathcal{M})}^{\text{posterior}} = \frac{ \overbrace{p(d|\theta, \mathcal{M})}^{\text{likelihood}} \ \overbrace{p(\theta|\mathcal{M})}^{\text{prior}}}{\underbrace{p(d|\mathcal{M})}_{\text{evidence}}}$$
- The prior is our belief about the model parameters before observing the data.
- The likelihood is the probability distribution of data given particular model parameters $\theta$.
- The posterior is the probability distribution of model parameters given particular data $d$.