- Jeffreys prior
In
Bayesian probability , the Jeffreys prior (called afterHarold Jeffreys ) is a
non-informativeprior distribution proportional to thesquare root of theFisher information ::
and is invariant under
reparameterization of .It's an important uninformative (objective) prior.
It allows us to describe our knowledge on , a transformation of with an improper uniform distribution. This also implies the resulting likelihood function, should be
asymptotically translated by changes in data. Due to asymptotical normality, this means only the first moment will vary when data is updated.It can be derived as follows:
We need an injective transformation of such that our prior under this transformation is uniform. It gives us "no information". We then use the following relation:
:
To conclude,
:
:
From a practical and mathematical standpoint, a valid reason to use this noninformative prior instead of others, like the ones obtained through a limit in conjugate families of distributions, is that it best represents the lack of knowledge when a certain
parametric family is chosen, and it is linked with strong Bayesian statistics results.In general, use of Jeffreys priors violates the
likelihood principle ; some statisticians therefore regard their use as unjustified.Fact|date=December 2007References
*cite journal
last= Jeffreys | first=H. | authorlink=Harold Jeffreys
year = 1946
title = An Invariant Form for the Prior Probability in Estimation Problems
journal = Proceedings of the Royal Society of London. Series A, Mathematical and Physical Sciences
volume = 186
issue = 1007
pages = 453–461
url = http://links.jstor.org/sici?sici=0080-4630(19460924)186%3A1007%3C453%3AAIFFTP%3E2.0.CO%3B2-J
doi = 10.1098/rspa.1946.0056*cite book
last= Jeffreys | first=H. | authorlink=Harold Jeffreys
year = 1939
title = Theory of Probability
publisher = Oxford University Press
Wikimedia Foundation. 2010.