Quote:
Originally Posted by BobSmith
[You must be logged in to view images. Log in or Register.]
Droodler, can you put that in layman's terms?
That last chart was clear, but I did not understand the rest.
|
Bayesian statistics involve three things: a prior distribution, a likelihood, and posterior (the result)
The prior distribution is what we expect the outcome to be before we apply the data. It is an assumption made from existing data or assumptions you make. Here, I made the assumption that 1 FBSS every 10 hours was reasonable. I made a prior based on this, that in 61 attempts you would get 1 FBSS.
Then what you do is take into account the actual data. Here, the OP said he had 109 spawns with 0 FBSS. I used that to create a likelihood (the red curve).
Then you combine them to get the posterior. That is the gold curve.
To interpret this graph
[You must be logged in to view images. Log in or Register.]
Think of the x axis as the probability of a single spawn dropping an FBSS. The Y axis is a density, so you can interpret the height of each curve at a point as the probability of that being the ACTUAL FBSS drop rate.
You can see that in the prior (blue curve), the expected FBSS drop rate is around 0.018. In the post, it is 0.012.
The cumulative function I think you understand.