hildensia/bayesian_changepoint_detection

Why the probability exceeds one?

mike-ocean opened this issue · 2 comments

I ran the given online detection example in the notebook, and I assumed the y axis indicating the probability of changepoint (am I right?). But the y value ranged from zero to hundreds.
I am not very familiar with the math, so can anyone please explain this outcome?

Thanks.

The third plot in the online example is the probability. The second plot is at each timestep a distribution over the length of the current part.

The third plot in the online example is the probability. The second plot is at each timestep a distribution over the length of the current part.

Thanks you Johannes! I realized this yesterday and just closed the issue. But thanks anyway!