I How to Express Non-regular Prior Distributions by Mathematical Formula

  • I
  • Thread starter Thread starter nizi
  • Start date Start date
  • Tags Tags
    Support
nizi
Messages
17
Reaction score
1
This time I target the following two-class Bayesian logistic regression as statistical models.

$$y_n \sim \mathrm{Bernoulli}(q_n)$$
$$q_n = \sigma (\beta_0 + x_n \beta_1)$$

where ##n## is the index of the data and ##\sigma## is the logistic function.

Since I assume ##\beta_0 + x_n \beta_1## as the linear predictor (the independent variable of the logistic activation function), I have two parameters ##\beta_0## and ##\beta_1##, and I want to express that their prior distributions independently follow a non-regular uniform distribution, the support of which each probability density function is the set of real numbers.
Is it appropriate to write

$$\beta_0, \beta_1 \sim \mathrm{Uniform}(-\infty, \infty), i.i.d.$$

in this case?
I'm particularly concerned about the appropriateness of writing i.i.d.
If there is a more appropriate way to express this, please let me know.
 
Last edited:
Physics news on Phys.org
The iid is fine. But this prior will require some justification. The uniform distribution will be really bad here because it has no tails so the posteriors are almost guaranteed to be unreliable. And starting with this prior is numerically impossible. And approximating it will probably have poor convergence.

If you do use this prior you will need to provide rock-solid theoretical arguments why the priors must come from a uniform distribution and none other. And if you have that much information then surely the non-regular version is not correct.
 
I appreciate your thought-provoking response very much.
My question has been answered.
What I was particularly concerned about is whether I can use the term "i.i.d." here, even thought ##x_n## has certain units, e.g. kg.
This is because the units of ##\beta_0## and ##\beta_1## are different in this case.

As you mentioned, it's impossible to use this improper prior in a numerical simulation, so it's only an mathematical expression.
 
Hi all, I've been a roulette player for more than 10 years (although I took time off here and there) and it's only now that I'm trying to understand the physics of the game. Basically my strategy in roulette is to divide the wheel roughly into two halves (let's call them A and B). My theory is that in roulette there will invariably be variance. In other words, if A comes up 5 times in a row, B will be due to come up soon. However I have been proven wrong many times, and I have seen some...
Thread 'Detail of Diagonalization Lemma'
The following is more or less taken from page 6 of C. Smorynski's "Self-Reference and Modal Logic". (Springer, 1985) (I couldn't get raised brackets to indicate codification (Gödel numbering), so I use a box. The overline is assigning a name. The detail I would like clarification on is in the second step in the last line, where we have an m-overlined, and we substitute the expression for m. Are we saying that the name of a coded term is the same as the coded term? Thanks in advance.

Similar threads

Replies
1
Views
2K
Replies
5
Views
4K
Replies
125
Views
19K
4
Replies
150
Views
19K
Replies
3
Views
2K
Replies
13
Views
3K
Replies
4
Views
4K
Back
Top