You do not have permission to edit this page, for the following reasons:

This IP address has been blocked from editing Wikipedia.
This does not affect your ability to read Wikipedia pages.
Most people who see this message have done nothing wrong. Some kinds of blocks restrict editing from specific service providers or telecom companies in response to recent abuse or vandalism, and can sometimes affect other users who are unrelated to that abuse. Review the information below for assistance if you do not believe that you have done anything wrong.

The IP address or range 3.89.0.0/16 has been blocked (disabled) by ‪L235‬ for the following reason(s):

The IP address that you are currently using has been blocked because it is believed to be a web host provider or colocation provider. To prevent abuse, web hosts and colocation providers may be blocked from editing Wikipedia.
You will not be able to edit Wikipedia using a web host or colocation provider because it hides your IP address, much like a proxy or VPN.
We recommend that you attempt to use another connection to edit. For example, if you use a proxy or VPN to connect to the internet, turn it off when editing Wikipedia. If you edit using a mobile connection, try using a Wi-Fi connection, and vice versa. If you are using a corporate internet connection, switch to a different Wi-Fi network. If you have a Wikipedia account, please log in.
If you do not have any other way to edit Wikipedia, you will need to request an IP block exemption.

How to appeal if you are confident that your connection does not use a colocation provider's IP address:
If you are confident that you are not using a web host, you may appeal this block by adding the following text on your talk page: ((unblock|reason=Caught by a colocation web host block but this host or IP is not a web host. My IP address is _______. Place any further information here. ~~~~)). You must fill in the blank with your IP address for this block to be investigated. Your IP address can be determined here. Alternatively, if you wish to keep your IP address private you can use the unblock ticket request system. There are several reasons you might be editing using the IP address of a web host or colocation provider (such as if you are using VPN software or a business network); please use this method of appeal only if you think your IP address is in fact not a web host or colocation provider.

Administrators: The IP block exemption user right should only be applied to allow users to edit using web host in exceptional circumstances, and requests should usually be directed to the functionaries team via email. If you intend to give the IPBE user right, a CheckUser needs to take a look at the account. This can be requested most easily at SPI Quick Checkuser Requests. Unblocking an IP or IP range with this template is highly discouraged without at least contacting the blocking administrator.

This block will expire on 18:20, 24 March 2026.

Even when blocked, you will usually still be able to edit your user talk page, as well as email administrators and other editors.

For information on how to proceed, please read the FAQ for blocked users and the guideline on block appeals. The guide to appealing blocks may also be helpful.

Other useful links: Blocking policy · Help:I have been blocked
Your IP address is in a range that has been blocked on all Wikimedia Foundation wikis. The block was made by ‪Wiki13‬. The reason given is Open proxy/Webhost: Visit the FAQ if you are affected .
- Start of block: 13:43, 5 May 2022
- Expiry of block: 13:43, 5 May 2024
Your current IP address is 3.89.215.189. The blocked range is 3.89.0.0/16. Please include all above details in any queries you make. If you believe you were blocked by mistake, you can find additional information and instructions in the No open proxies global policy. Otherwise, to discuss the block please post a request for review on Meta-Wiki. You could also send an email to the stewards VRT queue at stewards@wikimedia.org including all above details.

You can view and copy the source of this page:

((multiple issues|
((More footnotes|date = July 2010))
((refimprove|date = July 2010))
))

In [[statistics]], the concept of being an '''invariant estimator''' is a criterion that can be used to compare the properties of different [[estimator]]s for the same quantity. It is a way of formalising the idea that an estimator should have certain intuitively appealing qualities. Strictly speaking, "invariant" would mean that the estimates themselves are unchanged when both the measurements and the parameters are transformed in a compatible way, but the meaning has been extended to allow the estimates to change in appropriate ways with such transformations.<ref>see section 5.2.1 in Gourieroux, C. and Monfort, A. (1995). Statistics and econometric models, volume 1. Cambridge University Press.</ref> The term '''equivariant estimator''' is used in formal mathematical contexts that include a precise description of the relation of the way the estimator changes in response to changes to the dataset and parameterisation: this corresponds to the use of "[[Equivariant map|equivariance]]" in more general mathematics.

==General setting==

===Background===

In [[statistical inference]], there are several approaches to [[estimation theory]] that can be used to decide immediately what estimators should be used according to those approaches. For example, ideas from [[Bayesian inference]] would lead directly to [[Bayesian estimator]]s. Similarly, the theory of classical statistical inference can sometimes lead to strong conclusions about what estimator should be used. However, the usefulness of these theories depends on having a fully prescribed [[statistical model]] and may also depend on having a relevant loss function to determine the estimator. Thus a [[Bayesian inference|Bayesian analysis]] might be undertaken, leading to a posterior distribution for relevant parameters, but the use of a specific utility or loss function may be unclear. Ideas of invariance can then be applied to the task of summarising the posterior distribution. In other cases, statistical analyses are undertaken without a fully defined statistical model or the classical theory of statistical inference cannot be readily applied because the family of models being considered are not amenable to such treatment. In addition to these cases where general theory does not prescribe an estimator, the concept of invariance of an estimator can be applied when seeking estimators of alternative forms, either for the sake of simplicity of application of the estimator or so that the estimator is [[Robust statistics|robust]].

The concept of invariance is sometimes used on its own as a way of choosing between estimators, but this is not necessarily definitive. For example, a requirement of invariance may be incompatible with the requirement that the [[Bias of an estimator|estimator be mean-unbiased]]; on the other hand, the criterion of [[Median#Median-unbiased estimators|median-unbiasedness]] is defined in terms of the estimator's sampling distribution and so is invariant under many transformations.

One use of the concept of invariance is where a class or family of estimators is proposed and a particular formulation must be selected amongst these. One procedure is to impose relevant invariance properties and then to find the formulation within this class that has the best properties, leading to what is called the optimal invariant estimator.

===Some classes of invariant estimators===
There are several types of transformations that are usefully considered when dealing with invariant estimators. Each gives rise to a class of estimators which are invariant to those particular types of transformation.
*Shift invariance: Notionally, estimates of a [[location parameter]] should be invariant to simple shifts of the data values. If all data values are increased by a given amount, the estimate should change by the same amount. When considering estimation using a [[weighted average]], this invariance requirement immediately implies that the weights should sum to one. While the same result is often derived from a requirement for unbiasedness, the use of "invariance" does not require that a mean value exists and makes no use of any probability distribution at all.
*Scale invariance: Note that this topic about the invariance of the estimator scale parameter not to be confused with the more general  [[scale invariance]] about the behavior of systems under aggregate properties (in physics).
*Parameter-transformation invariance: Here, the transformation applies to the parameters alone. The concept here is that essentially the same inference should be made from data and a model involving a parameter θ as would be made from the same data if the model used a parameter φ, where φ is a one-to-one transformation of θ, φ=''h''(θ). According to this type of invariance, results from transformation-invariant estimators should also be related by φ=''h''(θ). [[Maximum likelihood estimator]]s have this property when the transformation is [[Monotonic function|monotonic]]. Though the asymptotic properties of the estimator might be invariant, the small sample properties can be different, and a specific distribution needs to be derived.<ref>Gouriéroux and Monfort (1995)</ref>
*Permutation invariance: Where a set of data values can be represented by a statistical model that they are outcomes from [[independent and identically distributed]] [[random variables]], it is reasonable to impose the requirement that any estimator of any property of the common distribution should be permutation-invariant: specifically that the estimator, considered as a function of the set of data-values, should not change if items of data are swapped within the dataset.

The combination of permutation invariance and location invariance for estimating a location parameter from an [[independent and identically distributed]] dataset using a weighted average implies that the weights should be identical and sum to one. Of course, estimators other than a weighted average may be preferable.

===Optimal invariant estimators===
Under this setting, we are given a set of measurements <math>x</math> which contains information about an unknown parameter <math>\theta</math>. The measurements <math>x</math> are modelled as a [[random vector|vector random variable]] having a [[probability density function]] <math>f(x|\theta)</math> which depends on  a parameter vector <math>\theta</math>.

The problem is to estimate <math>\theta</math> given <math>x</math>. The estimate, denoted by <math>a</math>, is a function of the measurements and belongs to a set <math>A</math>. The quality of the result is defined by a [[loss function]] <math>L=L(a,\theta)</math> which determines a [[risk function]] <math>R=R(a,\theta)=E[L(a,\theta)|\theta]</math>. The sets of possible values of <math>x</math>, <math>\theta</math>, and <math>a</math> are denoted by <math>X</math>, <math>\Theta</math>, and <math>A</math>, respectively.

===In classification===
In [[statistical classification]], the rule which assigns a class to a new data-item can be considered to be a special type of estimator. A number of invariance-type considerations can be brought to bear in formulating [[prior knowledge for pattern recognition]].

==Mathematical setting==

=== Definition ===
An invariant estimator is an estimator which obeys the following two rules:((citation needed|date=November 2010))
# Principle of Rational Invariance: The action taken in a decision problem should not depend on transformation on the measurement used
# Invariance Principle: If two decision problems have the same formal structure (in terms of <math>X</math>, <math>\Theta</math>, <math>f(x|\theta)</math> and <math>L</math>), then the same decision rule should be used in each problem.

To define an invariant or equivariant estimator formally, some definitions related to groups of transformations are needed first. Let <math>X</math> denote the set of possible data-samples. A [[group of transformation]]s of <math>X</math>, to be denoted by <math>G</math>, is a set of (measurable) 1:1 and onto transformations of <math>X</math> into itself, which satisfies the following conditions:

# If <math>g_1\in G</math>  and <math>g_2\in G</math> then <math>g_1 g_2\in G \,</math>
# If <math>g\in G</math> then <math>g^{-1}\in G</math>, where <math>g^{-1}(g(x))=x \, .</math> (That is, each transformation has an inverse within the group.)
# <math>e\in G</math> (i.e. there is an identity transformation <math> e(x)=x \, </math>)

Datasets <math>x_1</math> and <math>x_2</math> in <math>X</math> are equivalent if <math>x_1=g(x_2)</math> for some <math>g\in G</math>. All the equivalent points form an [[equivalence class]].
Such an equivalence class is called an [[orbit (group theory)|orbit]] (in <math>X</math>). The <math>x_0</math> orbit, <math>X(x_0)</math>, is the set <math>X(x_0)=\{g(x_0):g\in G\}</math>.
If <math>X</math> consists of a single orbit then <math>g</math> is said to be transitive.

A family of densities <math>F</math> is said to be invariant under the group <math>G</math> if, for every <math>g\in G</math> and <math>\theta\in \Theta</math> there exists a unique <math>\theta^*\in  \Theta</math> such that <math>Y=g(x)</math> has density <math>f(y|\theta^*)</math>. <math>\theta^*</math> will be denoted <math>\bar{g}(\theta)</math>.

If <math>F</math> is invariant under the group <math>G</math> then the loss function <math>L(\theta,a)</math> is said to be invariant under <math>G</math> if for every <math>g\in G</math> and <math>a\in A</math> there exists an <math>a^*\in A</math> such that <math>L(\theta,a)=L(\bar{g}(\theta),a^*)</math> for all <math>\theta \in \Theta</math>. The transformed value <math>a^*</math> will be denoted by <math>\tilde{g}(a)</math>.

In the above, <math>\bar{G}=\{\bar{g}:g\in G\}</math> is a group of transformations from <math>\Theta</math> to itself and <math>\tilde{G}=\{\tilde{g}: g \in G\}</math> is a group of transformations from <math>A</math> to itself.

An estimation problem is invariant(equivariant) under <math>G</math> if there exist three groups <math>G, \bar{G}, \tilde{G}</math> as defined above.

For an estimation problem that is invariant under <math>G</math>, estimator <math>\delta(x)</math> is an invariant estimator under <math>G</math> if, for all <math>x\in X</math> and <math>g\in G</math>,
:<math>\delta(g(x)) = \tilde{g}(\delta(x)).</math>

=== Properties ===

# The risk function of an invariant estimator, <math>\delta</math>, is constant on orbits of <math>\Theta</math>. Equivalently <math>R(\theta,\delta)=R(\bar{g}(\theta),\delta)</math> for all <math>\theta \in \Theta</math> and <math>\bar{g}\in \bar{G}</math>.
# The risk function of an invariant estimator with transitive <math>\bar{g}</math> is constant.

For a given problem, the invariant estimator with the lowest risk is termed the "best invariant estimator". Best invariant estimator cannot always be achieved. A special case for which it can be achieved is the case when <math>\bar{g}</math> is transitive.

=== Example: Location parameter ===

Suppose <math>\theta</math> is a location parameter if the density of <math>X</math> is of the form <math>f(x-\theta)</math>. For <math> \Theta=A=\mathbb{R}^1 </math> and <math>L=L(a-\theta)</math>, the problem is invariant under <math>g=\bar{g}=\tilde{g}=\{g_c:g_c(x)=x+c, c\in \mathbb{R}\}</math>. The invariant estimator in this case must satisfy
:<math>\delta(x+c)=\delta(x)+c, \text{ for all } c\in \mathbb{R},</math>
thus it is of the form <math>\delta(x)=x+K</math> (<math>K\in \mathbb{R}</math>). <math>\bar{g}</math> is transitive on <math>\Theta</math> so the risk does not vary with <math>\theta</math>: that is, <math>R(\theta,\delta)=R(0,\delta)=\operatorname{E}[L(X+K)|\theta=0]</math>. The best invariant estimator is the one that brings the risk <math>R(\theta,\delta)</math> to minimum.

In the case that L is the squared error <math>\delta(x)=x-\operatorname{E}[X|\theta=0].</math>

=== Pitman estimator ===
The estimation problem is that <math>X=(X_1,\dots,X_n)</math> has density <math>f(x_1-\theta,\dots,x_n-\theta)</math>, where ''&theta;'' is a parameter to be estimated, and where the [[loss function]] is <math>L(|a-\theta|)</math>. This problem is invariant with the following (additive) transformation groups:
:<math>G=\{g_c:g_c(x)=(x_1+c, \dots, x_n+c),c\in \mathbb{R}^1\},</math>
:<math>\bar{G}=\{g_c:g_c(\theta)=\theta + c,c\in \mathbb{R}^1\},</math>
:<math>\tilde{G}=\{g_c:g_c(a)=a + c,c\in \mathbb{R}^1\} .</math>

The best invariant estimator <math>\delta(x)</math> is the one that minimizes
:<math>\frac{\int_{-\infty}^{\infty} L(\delta(x)-\theta)f(x_1-\theta,\dots,x_n-\theta)d\theta}{\int_{-\infty}^{\infty}f(x_1-\theta,\dots,x_n-\theta)d\theta},</math>
and this is Pitman's estimator (1939).

For the squared error loss case, the result is
:<math>\delta(x)=\frac{\int_{-\infty}^{\infty} \theta f(x_1-\theta,\dots,x_n-\theta)d\theta}{\int_{-\infty}^{\infty}f(x_1-\theta,\dots,x_n-\theta)d\theta}.</math>

If <math>x \sim N(\theta 1_n,I)\,\!</math> (i.e. a [[multivariate normal distribution]] with independent, unit-variance components) then
:<math>\delta_{\text{Pitman)) = \delta_{ML}=\frac{\sum{x_i)){n}.</math>

If <math>x \sim C(\theta 1_n,I \sigma^2)\,\!</math> (independent components having a [[Cauchy distribution]] with scale parameter ''&sigma;'') then
<math>\delta_{\text{Pitman)) \ne \delta_{ML}</math>,. However the result is
:<math>\delta_{\text{Pitman))=\sum_{k=1}^n{x_k\left[\frac{\text{Re}\{w_k\)){\sum_{m=1}^{n}{\text{Re}\{w_k\))}\right]}, \qquad n>1,</math>
with
:<math>w_k = \prod_{j\ne k}\left[\frac{1}{(x_k-x_j)^2+4\sigma^2}\right]\left[1-\frac{2\sigma}{(x_k-x_j)}i\right].</math>

==References==
((Reflist))
* ((cite book
 |title=Statistical decision theory and Bayesian Analysis
 |first=James O. |last=Berger |author-link=James Berger (statistician)
 |year=1985
 |edition=2nd
 |publisher=Springer-Verlag |location=New York
 |isbn=0-387-96098-8 |mr=0804611
))((Page needed|date=January 2011))
* ((cite journal | last1 = Freue | first1 = Gabriela V. Cohen | year = 2007 | title = The Pitman estimator of the Cauchy location parameter | journal = Journal of Statistical Planning and Inference | volume = 137 | issue = 6 | pages = 1900–1913 | doi = 10.1016/j.jspi.2006.05.002 ))
* ((cite journal | last1 = Pitman | first1 = E.J.G. | year = 1939 | title = The estimation of the location and scale parameters of a continuous population of any given form | journal = [[Biometrika]] | volume = 30 | issue = 3/4| pages = 391–421 | jstor = 2332656 | doi=10.1093/biomet/30.3-4.391))
* ((cite journal | last1 = Pitman | first1 = E.J.G. | year = 1939 | title = Tests of Hypotheses Concerning Location and Scale Parameters | journal = [[Biometrika]] | volume = 31 | issue = 1/2| pages = 200–215 | jstor = 2334983 | doi=10.1093/biomet/31.1-2.200))

((DEFAULTSORT:Invariant Estimator))
[[Category:Estimator]]
[[Category:Invariant theory]]

Pages transcluded onto the current version of this page (help):

Invariant estimator (edit)
Template:Ambox (view source) (template editor protected)
Template:Category handler (view source) (protected)
Template:Citation needed (view source) (protected)
Template:Cite book (view source) (protected)
Template:Cite journal (view source) (protected)
Template:Delink (view source) (protected)
Template:Find sources mainspace (view source) (template editor protected)
Template:Fix (view source) (protected)
Template:Fix/category (view source) (protected)
Template:Main other (view source) (protected)
Template:More citations needed (view source) (template editor protected)
Template:More footnotes (view source) (template editor protected)
Template:More footnotes needed (view source) (template editor protected)
Template:Multiple issues (view source) (template editor protected)
Template:Multiple issues/styles.css (view source) (template editor protected)
Template:Page needed (view source) (template editor protected)
Template:Refimprove (view source) (template editor protected)
Template:Reflist (view source) (protected)
Template:Reflist/styles.css (view source) (protected)
Template:Yesno (view source) (protected)
Template:Yesno-no (view source) (template editor protected)
Module:Arguments (view source) (protected)
Module:Category handler (view source) (protected)
Module:Category handler/blacklist (view source) (protected)
Module:Category handler/config (view source) (protected)
Module:Category handler/data (view source) (protected)
Module:Category handler/shared (view source) (protected)
Module:Check for unknown parameters (view source) (protected)
Module:Citation/CS1 (view source) (protected)
Module:Citation/CS1/COinS (view source) (protected)
Module:Citation/CS1/Configuration (view source) (protected)
Module:Citation/CS1/Date validation (view source) (protected)
Module:Citation/CS1/Identifiers (view source) (protected)
Module:Citation/CS1/Utilities (view source) (protected)
Module:Citation/CS1/Whitelist (view source) (protected)
Module:Citation/CS1/styles.css (view source) (protected)
Module:Delink (view source) (protected)
Module:Find sources (view source) (template editor protected)
Module:Find sources/config (view source) (template editor protected)
Module:Find sources/links (view source) (template editor protected)
Module:Find sources/templates/Find sources mainspace (view source) (template editor protected)
Module:Message box (view source) (protected)
Module:Message box/ambox.css (view source) (protected)
Module:Message box/configuration (view source) (protected)
Module:Namespace detect/config (view source) (protected)
Module:Namespace detect/data (view source) (protected)
Module:String (view source) (protected)
Module:Unsubst (view source) (protected)
Module:Yesno (view source) (protected)

Return to Invariant estimator.

Retrieved from "https://en.wikipedia.org/wiki/Invariant_estimator"