In summary, the Bayesian interpretation of probability is more appropriate for some situations than the frequentist interpretation, and the informed prior can produce more accurate results than the default prior.
  • #1
35,204
13,431
Confessions of a moderate Bayesian, part 1
Bayesian statistics by and for non-statisticians
https://www.cafepress.com/physicsforums.13265286
Background
I am a statistics enthusiast, although I am not a statistician. My training is in biomedical engineering, and I have been heavily involved in the research and development of novel medical imaging technologies for the bulk of my career. Due to my enjoyment of statistics and my professional research needs I was usually the author in charge of the statistical analysis of most of our papers.
In the course of that work, I developed an interest in and an appreciation of Bayesian statistics, which I have used in the analysis of most of my more recent projects. I have published the Bayesian analysis of some of these studies, although more were done in addition to standard (published) frequentist analysis.
This article will be a quick and practical how-to on getting started doing...

Continue reading...
 
Last edited by a moderator:
  • Like
Likes pinball1970, WWGD, atyy and 8 others
Physics news on Phys.org
  • #2
Nice article, @Dale!

Bayesian thinking has also been interesting and confusing to me. I read about it and understand it until I try to explain it and then I just don't have the proper intuition to understand its deeper meaning.
 
  • Like
Likes Dale
  • #3
I plan on following up with some more articles that will deal with Bayesian probability itself and the use of Bayesian statistics in science. I agree that it can be a little hard at first too.
 
  • #4
Although I took a course as an undergraduate ostensibly covering Bayesian models, I realize I learned approximately nothing! (I believe it was in the psych dept...part of rounding out my bachelor of arts)
That being said I have learned much from your various explanations and look forward to the article(s). Thanks.
 
  • Like
Likes WWGD, Greg Bernhardt and Dale
  • #5
hutchphd said:
That being said I have learned much from your various explanations and look forward to the article(s).
I appreciate that! Hopefully I can get the next one fleshed out tonight or tomorrow.
 
  • #6
hutchphd said:
Although I took a course as an undergraduate ostensibly covering Bayesian models, I realize I learned approximately nothing! (I believe it was in the psych dept...part of rounding out my bachelor of arts)
That being said I have learned much from your various explanations and look forward to the article(s). Thanks.
That may just be your prior. Let's ask you a few questions to test your knowledge to get a posterior, like I did when I asked to show full work in an exam, given a Prior: " Show your Posterior" . Won't make that mistake again ;)
 
  • Like
Likes hutchphd
  • #7
  • #8
WWGD said:
Link points to the PF Shop, not the Insights .
It seems to be working now, but I didn’t do anything to fix it
 
  • #9
May just be my PC. This is what I am getting:

1606971048548.png
 
  • #10
Never mind. I had run out of my monthly data allotment and I get weird stuff like this occasionally.

EDIT: Have you used R within Jupyter? Jupyter stands for : Julya, Python and R. Now too, SQL Server Developer 2017 has both R and a Python Servers.
 
  • #12
WWGD said:
Never mind. I had run out of my monthly data allotment and I get weird stuff like this occasionally.

EDIT: Have you used R within Jupyter? Jupyter stands for : Julya, Python and R. Now too, SQL Server Developer 2017 has both R and a Python Servers.
I have not. I tried to get Jupyter set up because it sounded useful to have both Python and R together and I was curious about Julia. But my installation didn't work the first time around so I didn't pursue it further.
 
  • Like
Likes WWGD
  • #13
Dale said:
Hi

I recently ran several analyses using the informed prior vs the default prior. I understand why the same analyses using the informed prior gives a bigger Bayes factor compared to the default, but I didn’t expect the effect sizes to all be smaller (and credible intervals narrower) when using the informed prior. Does anybody know why this might be?
 

Similar threads

  • Set Theory, Logic, Probability, Statistics
Replies
7
Views
2K
  • Sticky
  • Set Theory, Logic, Probability, Statistics
Replies
13
Views
3K
  • Sticky
  • Set Theory, Logic, Probability, Statistics
2
Replies
44
Views
7K
  • Set Theory, Logic, Probability, Statistics
Replies
6
Views
1K
  • Set Theory, Logic, Probability, Statistics
Replies
3
Views
719
Replies
4
Views
2K
  • Set Theory, Logic, Probability, Statistics
Replies
17
Views
2K
  • STEM Academic Advising
Replies
6
Views
1K
  • STEM Academic Advising
Replies
1
Views
940
  • STEM Academic Advising
Replies
1
Views
970
Back
Top