# A little help understanding this bayesian problem (very basic)

1. Dec 20, 2011

### bennyska

1. The problem statement, all variables and given/known data
so i'm trying to teach myself bayes, and i got a book, and i'm going through trying to do the exercises, and lo and behold, i get stuck on the first one. i thought i was getting it, but the answer given at the back of the book is different than mine.

2. Relevant equations

3. The attempt at a solution
i've attached three pictures. the first is the actual problem. so when i did it myself, i got that the first column of priors should all be 1/10, since we're assuming that the urn can contain 0 up to 9 red balls, for a total of 10 possibilities. the second picture is the answer, where it seems that i should have said 1/9 for the first column. the third picture is an example that seems to me to indicate that it should be 1/10. what am i missing?

#### Attached Files:

File size:
35.7 KB
Views:
68
File size:
19.1 KB
Views:
67
• ###### bayes - prob1.jpg
File size:
24 KB
Views:
67
2. Dec 20, 2011

### Simon Bridge

There are 10 possible values for X, but there are only 9 balls to choose from.
What does the prior probability represent in this case?

3. Dec 20, 2011

### bennyska

the prior probability represents the number of red balls in the urn. i'm told to assume that each possibility is equally likely. so there might be zero balls, 1 ball, 2 balls, etc, up to 9 balls. that should be a 1/10 probability for each case. furthermore, shouldn't the prior column sum up to 1? in the answer, doesn't the prior column sum up to 10/9?

4. Dec 20, 2011

### Ray Vickson

You are correct. Either the "zero balls" case should not be present (giving 1/9 probability of each of the others) or else it is present and all priors should be 1/10.

RGV

5. Dec 20, 2011

### bennyska

thank you. also, i should have been more clear, the third picture was an example with the exact same problem, just fewer balls.

6. Dec 20, 2011

### Simon Bridge

Yes - I'm inclined to agree that the text is mistaken in the model answer.
Frustrating, I know.

I had a go trying to identify the mistake but failed. Why would the author think to assume that P(X=x)=x/(N-1)?

I'd ignore the answers but do the analysis anyway. Pause at each step to see if what you get makes sense. The other one seems to be right - maybe you want to start with that instead.