Equirectangular projection: Difference between revisions

From formulasearchengine
Jump to navigation Jump to search
Newer edit →
en>Luckas-bot
m r2.7.1) (Robot: Adding eo:Samdistanca cilindra projekcio
 
en>Strebe
Uh, yes? —could argue either way. Undid revision 590228252 by Strebe (talk)
Line 1: Line 1:
They contact me Emilia. For a while she's been in South Dakota. Managing people is what I do and the wage has been truly satisfying. Doing ceramics is what adore doing.<br><br>Check out my website [http://qmm.is/mealdeliveryservice77873 weight loss food programs]
<!-- The \,\! are to keep the formulas rendered as PNG instead of HTML. Please don't remove them. -->
In [[statistical decision theory]], an '''admissible decision rule''' is a rule for making a decision such that there is not any other rule that is always "better" than it.<ref>[[Yadolah Dodge|Dodge, Y.]] (2003) ''The Oxford Dictionary of Statistical Terms''. OUP. ISBN 0-19-920613-9 (entry for admissible decision function)</ref> 
 
Generally speaking, in most decision problems the set of admissible rules is large, even infinite, so this is not a sufficient criterion to pin down a single rule, but as will be seen there are some good reasons to favor admissible rules; compare [[Pareto efficiency]].
 
==Definition==
Define [[Set (mathematics)|sets]] <math>\Theta\,</math>, <math>\mathcal{X}</math> and <math>\mathcal{A}</math>, where <math>\Theta\,</math> are the states of nature, <math>\mathcal{X}</math> the possible observations, and <math>\mathcal{A}</math> the actions that may be taken. An observation <math>x \in \mathcal{X}\,\!</math> is distributed as <math>F(x\mid\theta)\,\!</math> and therefore provides evidence about the state of nature <math>\theta\in\Theta\,\!</math>.  A '''decision rule''' is a [[Function (mathematics)|function]] <math>\delta:{\mathcal{X}}\rightarrow {\mathcal{A}}</math>, where upon observing <math>x\in \mathcal{X}</math>, we choose to take action <math>\delta(x)\in \mathcal{A}\,\!</math>.
 
Also define a '''[[loss function]]''' <math>L: \Theta \times \mathcal{A} \rightarrow \mathbb{R}</math>, which specifies the loss we would incur by taking action <math>a \in \mathcal{A}</math> when the true state of nature is <math>\theta \in \Theta</math>. Usually we will take this action after observing data <math>x \in \mathcal{X}</math>, so that the loss will be <math>L(\theta,\delta(x))\,\!</math>.  (It is possible though unconventional to recast the following definitions in terms of a [[utility function]], which is the negative of the loss.)
 
Define the '''[[risk function]]''' as the [[expected value|expectation]]
 
:<math>R(\theta,\delta)=\operatorname{E}_{F(x\mid\theta)}[{L(\theta,\delta(x))]}.\,\!</math>
 
Whether a decision rule <math>\delta\,\!</math> has low risk depends on the true state of nature <math>\theta\,\!</math>.  A decision rule <math>\delta^*\,\!</math> '''[[dominating decision rule|dominates]]''' a decision rule <math>\delta\,\!</math> if and only if <math>R(\theta,\delta^*)\le R(\theta,\delta)</math> for all <math>\theta\,\!</math>, ''and'' the inequality is [[inequality (mathematics)|strict]] for some <math>\theta\,\!</math>.
 
A decision rule is '''admissible''' (with respect to the loss function) if and only if no other rule dominates it; otherwise it is '''inadmissible'''. Thus an admissible decision rule is a [[maximal element]] with respect to the above partial order.
An inadmissible rule is not preferred (except for reasons of simplicity or computational efficiency), since by definition there is some other rule that will achieve equal or lower risk for ''all'' <math>\theta\,\!</math>.  But just because a rule <math>\delta\,\!</math> is admissible does not mean it is a good rule to use.  Being admissible means there is no other single rule that is ''always'' better - but other admissible rules might achieve lower risk for most <math>\theta\,\!</math> that occur in practice.  (The Bayes risk discussed below is a way of explicitly considering which <math>\theta\,\!</math> occur in practice.)
 
==Bayes rules and generalized Bayes rules==
{{See also|Bayes estimator#Admissibility}}
 
===Bayes rules===
Let <math>\pi(\theta)\,\!</math> be a probability distribution on the states of nature. From a [[Bayesian probability|Bayesian]] point of view, we would regard it as a ''[[prior distribution]]''.  That is, it is our believed probability distribution on the states of nature, prior to observing data. For a [[Frequency probability|frequentist]], it is merely a function on <math>\Theta\,\!</math> with no such special interpretation. The '''Bayes risk''' of the decision rule <math>\delta\,\!</math> with respect to <math>\pi(\theta)\,\!</math> is the expectation
 
:<math>r(\pi,\delta)=\operatorname{E}_{\pi(\theta)}[R(\theta,\delta)].\,\!</math>
 
A decision rule <math>\delta\,\!</math> that minimizes <math>r(\pi,\delta)\,\!</math> is called a '''[[Bayes estimator|Bayes rule]]''' with respect to  <math>\pi(\theta)\,\!</math>.  There may be more than one such Bayes rule.  If the Bayes risk is infinite for all <math>\delta\,\!</math>, then no Bayes rule is defined.
 
===Generalized Bayes rules===
{{See also|Bayes_estimator#Generalized_Bayes_estimators}}
 
In the Bayesian approach to decision theory, the observed <math>x\,\!</math> is considered ''fixed''. Whereas the frequentist approach (i.e., risk) averages over possible samples <math>x \in \mathcal{X}\,\!</math>, the Bayesian would fix the observed sample <math>x\,\!</math> and average over hypotheses <math>\theta \in \Theta\,\!</math>. Thus, the Bayesian approach is to consider for our observed <math>x\,\!</math> the '''[[Loss_function#Expected_loss|expected loss]]'''
 
:<math>\rho(\pi,\delta \mid x)=\operatorname{E}_{\pi(\theta \mid x)} [ L(\theta,\delta(x)) ]. \,\!</math>
 
where the expectation is over the ''posterior'' of <math>\theta\,\!</math> given <math>x\,\!</math> (obtained from <math>\pi(\theta)\,\!</math> and <math>F(x\mid\theta)\,\!</math> using [[Bayes' theorem]]).
 
Having made explicit the expected loss for each given <math>x\,\!</math> separately, we can define a decision rule <math>\delta\,\!</math> by specifying for each <math>x\,\!</math> an action <math>\delta(x)\,\!</math> that minimizes the expected loss.  This is known as a '''generalized Bayes rule''' with respect to <math>\pi(\theta)\,\!</math>.  There may be more than one generalized Bayes rule, since there may be multiple choices of <math>\delta(x)\,\!</math> that achieve the same expected loss.
 
At first, this may appear rather different from the Bayes rule approach of the previous section, not a generalization.  However, notice that the Bayes risk already averages over <math>\Theta\,\!</math> in Bayesian fashion, and the Bayes risk may be recovered as the expectation over <math>\mathcal{X}</math> of the expected loss (where <math>x\sim\theta\,\!</math> and <math>\theta\sim\pi\,\!</math>).  Roughly speaking, <math>\delta\,\!</math> minimizes this expectation of expected loss (i.e., is a Bayes rule) if it minimizes the expected loss for each <math>x \in \mathcal{X}</math> separately (i.e., is a generalized Bayes rule).
 
Then why is the notion of generalized Bayes rule an improvement?  It is indeed equivalent to the notion of Bayes rule when a Bayes rule exists and all <math>x\,\!</math> have positive probability.  However, no Bayes rule exists if the Bayes risk is infinite (for all <math>\delta\,\!</math>).  In this case it is still useful to define a generalized Bayes rule <math>\delta\,\!</math>, which at least chooses a minimum-expected-loss action <math>\delta(x)\!\,</math> for those <math>x\,\!</math> for which a finite-expected-loss action does exist.  In addition, a generalized Bayes rule may be desirable because it must choose a minimum-expected-loss action <math>\delta(x)\,\!</math> for ''every'' <math>x\,\!</math>, whereas a Bayes rule would be allowed to deviate from this policy on a set <math>X \subseteq \mathcal{X}</math> of measure 0 without affecting the Bayes risk.
 
More important, it is sometimes convenient to use an improper prior <math>\pi(\theta)\,\!</math>.  In this case, the Bayes risk is not even well-defined, nor is there any well-defined distribution over <math>x\,\!</math>.  However, the posterior <math>\pi(\theta\mid x)\,\!</math>—and hence the expected loss—may be well-defined for each <math>x\,\!</math>, so that it is still possible to define a generalized Bayes rule.
 
===Admissibility of (generalized) Bayes rules===
According to the complete class theorems, under mild conditions every admissible rule is a (generalized) Bayes rule (with respect to some prior <math>\pi(\theta)\,\!</math>—possibly an improper one—that favors distributions <math>\theta\,\!</math> where that rule achieves low risk). Thus, in [[frequentist]] [[decision theory]] it is sufficient to consider only (generalized) Bayes rules.
 
Conversely, while Bayes rules with respect to proper priors are virtually always admissible, generalized Bayes rules corresponding to [[Prior probability#Improper priors|improper priors]] need not yield admissible procedures. [[Stein's example]] is one such famous situation.
 
==Examples==
 
The [[James–Stein estimator]] is a nonlinear estimator which can be shown to dominate, or outperform, the [[ordinary least squares]] technique with respect to a mean-square error loss function.<ref>{{harvnb|Cox|Hinkley|1974|loc=Section 11.8}}</ref> Thus least squares estimation is not necessarily an admissible estimation procedure. Some others of the standard estimates associated with the [[normal distribution]] are also inadmissible: for example, the [[sample variance|sample estimate of the variance]] when the population mean and variance are unknown.<ref>{{harvnb|Cox|Hinkley|1974|loc=Exercise 11.7}}</ref>
 
==See also==
*[[Decision theory]]
*[[Maximal element]]
*[[Pareto efficiency]]
 
{{More footnotes|date=July 2010}}
 
==Notes==
{{Reflist}}
 
==References==
*{{cite book |last1=Cox |first1=D. R. |last2=Hinkley |first2=D. V. |title=Theoretical Statistics |publisher=Wiley |year=1974 |isbn=0-412-12420-3 |ref=harv }}
*{{cite book |last=Berger |first=James O. |title=Statistical Decision Theory and Bayesian Analysis |publisher=Springer-Verlag |location= |year=1980 |isbn=0-387-96098-8 |edition=2nd }}
*{{cite book |author=DeGroot, Morris |authorlink=Morris DeGroot |title=Optimal Statistical Decisions |publisher=Wiley Classics Library |location= |year=2004 |isbn=0-471-68029-X |origyear=1st. pub. 1970 }}
*{{cite book |author=Robert, Christian P. |title=The Bayesian Choice |publisher=Springer-Verlag |location= |year=1994 |isbn=3-540-94296-3 }}
 
[[Category:Bayesian statistics]]
[[Category:Decision theory]]

Revision as of 23:26, 11 January 2014

In statistical decision theory, an admissible decision rule is a rule for making a decision such that there is not any other rule that is always "better" than it.[1]

Generally speaking, in most decision problems the set of admissible rules is large, even infinite, so this is not a sufficient criterion to pin down a single rule, but as will be seen there are some good reasons to favor admissible rules; compare Pareto efficiency.

Definition

Define sets Θ, 𝒳 and 𝒜, where Θ are the states of nature, 𝒳 the possible observations, and 𝒜 the actions that may be taken. An observation x∈𝒳 is distributed as F(x∣θ) and therefore provides evidence about the state of nature θ∈Θ. A decision rule is a function δ:𝒳→𝒜, where upon observing x∈𝒳, we choose to take action δ(x)∈𝒜.

Also define a loss function L:Θ×𝒜→ℝ, which specifies the loss we would incur by taking action a∈𝒜 when the true state of nature is θ∈Θ. Usually we will take this action after observing data x∈𝒳, so that the loss will be L(θ,δ(x)). (It is possible though unconventional to recast the following definitions in terms of a utility function, which is the negative of the loss.)

Define the risk function as the expectation

R(θ,δ)=EF(x∣θ)[L(θ,δ(x))].

Whether a decision rule δ has low risk depends on the true state of nature θ. A decision rule δ* dominates a decision rule δ if and only if R(θ,δ*)≤R(θ,δ) for all θ, and the inequality is strict for some θ.

A decision rule is admissible (with respect to the loss function) if and only if no other rule dominates it; otherwise it is inadmissible. Thus an admissible decision rule is a maximal element with respect to the above partial order. An inadmissible rule is not preferred (except for reasons of simplicity or computational efficiency), since by definition there is some other rule that will achieve equal or lower risk for all θ. But just because a rule δ is admissible does not mean it is a good rule to use. Being admissible means there is no other single rule that is always better - but other admissible rules might achieve lower risk for most θ that occur in practice. (The Bayes risk discussed below is a way of explicitly considering which θ occur in practice.)

Bayes rules and generalized Bayes rules

DTZ's public sale group in Singapore auctions all forms of residential, workplace and retail properties, outlets, homes, lodges, boarding homes, industrial buildings and development websites. Auctions are at present held as soon as a month.

We will not only get you a property at a rock-backside price but also in an space that you've got longed for. You simply must chill out back after giving us the accountability. We will assure you 100% satisfaction. Since we now have been working in the Singapore actual property market for a very long time, we know the place you may get the best property at the right price. You will also be extremely benefited by choosing us, as we may even let you know about the precise time to invest in the Singapore actual property market.

The Hexacube is offering new ec launch singapore business property for sale Singapore investors want to contemplate. Residents of the realm will likely appreciate that they'll customize the business area that they wish to purchase as properly. This venture represents one of the crucial expansive buildings offered in Singapore up to now. Many investors will possible want to try how they will customise the property that they do determine to buy by means of here. This location has offered folks the prospect that they should understand extra about how this course of can work as well.

Singapore has been beckoning to traders ever since the value of properties in Singapore started sky rocketing just a few years again. Many businesses have their places of work in Singapore and prefer to own their own workplace area within the country once they decide to have a everlasting office. Rentals in Singapore in the corporate sector can make sense for some time until a business has discovered a agency footing. Finding Commercial Property Singapore takes a variety of time and effort but might be very rewarding in the long term.

is changing into a rising pattern among Singaporeans as the standard of living is increasing over time and more Singaporeans have abundance of capital to invest on properties. Investing in the personal properties in Singapore I would like to applaud you for arising with such a book which covers the secrets and techniques and tips of among the profitable Singapore property buyers. I believe many novice investors will profit quite a bit from studying and making use of some of the tips shared by the gurus." – Woo Chee Hoe Special bonus for consumers of Secrets of Singapore Property Gurus Actually, I can't consider one other resource on the market that teaches you all the points above about Singapore property at such a low value. Can you? Condominium For Sale (D09) – Yong An Park For Lease

In 12 months 2013, c ommercial retails, shoebox residences and mass market properties continued to be the celebrities of the property market. Models are snapped up in report time and at document breaking prices. Builders are having fun with overwhelming demand and patrons need more. We feel that these segments of the property market are booming is a repercussion of the property cooling measures no.6 and no. 7. With additional buyer's stamp responsibility imposed on residential properties, buyers change their focus to commercial and industrial properties. I imagine every property purchasers need their property funding to understand in value.

Bayes rules

Let π(θ) be a probability distribution on the states of nature. From a Bayesian point of view, we would regard it as a prior distribution. That is, it is our believed probability distribution on the states of nature, prior to observing data. For a frequentist, it is merely a function on Θ with no such special interpretation. The Bayes risk of the decision rule δ with respect to π(θ) is the expectation

r(π,δ)=Eπ(θ)[R(θ,δ)].

A decision rule δ that minimizes r(π,δ) is called a Bayes rule with respect to π(θ). There may be more than one such Bayes rule. If the Bayes risk is infinite for all δ, then no Bayes rule is defined.

Generalized Bayes rules

DTZ's public sale group in Singapore auctions all forms of residential, workplace and retail properties, outlets, homes, lodges, boarding homes, industrial buildings and development websites. Auctions are at present held as soon as a month.

We will not only get you a property at a rock-backside price but also in an space that you've got longed for. You simply must chill out back after giving us the accountability. We will assure you 100% satisfaction. Since we now have been working in the Singapore actual property market for a very long time, we know the place you may get the best property at the right price. You will also be extremely benefited by choosing us, as we may even let you know about the precise time to invest in the Singapore actual property market.

The Hexacube is offering new ec launch singapore business property for sale Singapore investors want to contemplate. Residents of the realm will likely appreciate that they'll customize the business area that they wish to purchase as properly. This venture represents one of the crucial expansive buildings offered in Singapore up to now. Many investors will possible want to try how they will customise the property that they do determine to buy by means of here. This location has offered folks the prospect that they should understand extra about how this course of can work as well.

Singapore has been beckoning to traders ever since the value of properties in Singapore started sky rocketing just a few years again. Many businesses have their places of work in Singapore and prefer to own their own workplace area within the country once they decide to have a everlasting office. Rentals in Singapore in the corporate sector can make sense for some time until a business has discovered a agency footing. Finding Commercial Property Singapore takes a variety of time and effort but might be very rewarding in the long term.

is changing into a rising pattern among Singaporeans as the standard of living is increasing over time and more Singaporeans have abundance of capital to invest on properties. Investing in the personal properties in Singapore I would like to applaud you for arising with such a book which covers the secrets and techniques and tips of among the profitable Singapore property buyers. I believe many novice investors will profit quite a bit from studying and making use of some of the tips shared by the gurus." – Woo Chee Hoe Special bonus for consumers of Secrets of Singapore Property Gurus Actually, I can't consider one other resource on the market that teaches you all the points above about Singapore property at such a low value. Can you? Condominium For Sale (D09) – Yong An Park For Lease

In 12 months 2013, c ommercial retails, shoebox residences and mass market properties continued to be the celebrities of the property market. Models are snapped up in report time and at document breaking prices. Builders are having fun with overwhelming demand and patrons need more. We feel that these segments of the property market are booming is a repercussion of the property cooling measures no.6 and no. 7. With additional buyer's stamp responsibility imposed on residential properties, buyers change their focus to commercial and industrial properties. I imagine every property purchasers need their property funding to understand in value.

In the Bayesian approach to decision theory, the observed x is considered fixed. Whereas the frequentist approach (i.e., risk) averages over possible samples x∈𝒳, the Bayesian would fix the observed sample x and average over hypotheses θ∈Θ. Thus, the Bayesian approach is to consider for our observed x the expected loss

ρ(π,δ∣x)=Eπ(θ∣x)[L(θ,δ(x))].

where the expectation is over the posterior of θ given x (obtained from π(θ) and F(x∣θ) using Bayes' theorem).

Having made explicit the expected loss for each given x separately, we can define a decision rule δ by specifying for each x an action δ(x) that minimizes the expected loss. This is known as a generalized Bayes rule with respect to π(θ). There may be more than one generalized Bayes rule, since there may be multiple choices of δ(x) that achieve the same expected loss.

At first, this may appear rather different from the Bayes rule approach of the previous section, not a generalization. However, notice that the Bayes risk already averages over Θ in Bayesian fashion, and the Bayes risk may be recovered as the expectation over 𝒳 of the expected loss (where x∼θ and θ∼π). Roughly speaking, δ minimizes this expectation of expected loss (i.e., is a Bayes rule) if it minimizes the expected loss for each x∈𝒳 separately (i.e., is a generalized Bayes rule).

Then why is the notion of generalized Bayes rule an improvement? It is indeed equivalent to the notion of Bayes rule when a Bayes rule exists and all x have positive probability. However, no Bayes rule exists if the Bayes risk is infinite (for all δ). In this case it is still useful to define a generalized Bayes rule δ, which at least chooses a minimum-expected-loss action δ(x) for those x for which a finite-expected-loss action does exist. In addition, a generalized Bayes rule may be desirable because it must choose a minimum-expected-loss action δ(x) for every x, whereas a Bayes rule would be allowed to deviate from this policy on a set X⊆𝒳 of measure 0 without affecting the Bayes risk.

More important, it is sometimes convenient to use an improper prior π(θ). In this case, the Bayes risk is not even well-defined, nor is there any well-defined distribution over x. However, the posterior π(θ∣x)—and hence the expected loss—may be well-defined for each x, so that it is still possible to define a generalized Bayes rule.

Admissibility of (generalized) Bayes rules

According to the complete class theorems, under mild conditions every admissible rule is a (generalized) Bayes rule (with respect to some prior π(θ)—possibly an improper one—that favors distributions θ where that rule achieves low risk). Thus, in frequentist decision theory it is sufficient to consider only (generalized) Bayes rules.

Conversely, while Bayes rules with respect to proper priors are virtually always admissible, generalized Bayes rules corresponding to improper priors need not yield admissible procedures. Stein's example is one such famous situation.

Examples

The James–Stein estimator is a nonlinear estimator which can be shown to dominate, or outperform, the ordinary least squares technique with respect to a mean-square error loss function.[2] Thus least squares estimation is not necessarily an admissible estimation procedure. Some others of the standard estimates associated with the normal distribution are also inadmissible: for example, the sample estimate of the variance when the population mean and variance are unknown.[3]

See also

Template:More footnotes

Notes

43 year old Petroleum Engineer Harry from Deep River, usually spends time with hobbies and interests like renting movies, property developers in singapore new condominium and vehicle racing. Constantly enjoys going to destinations like Camino Real de Tierra Adentro.

References

  • 20 year-old Real Estate Agent Rusty from Saint-Paul, has hobbies and interests which includes monopoly, property developers in singapore and poker. Will soon undertake a contiki trip that may include going to the Lower Valley of the Omo.

    My blog: http://www.primaboinca.com/view_profile.php?userid=5889534
  • 20 year-old Real Estate Agent Rusty from Saint-Paul, has hobbies and interests which includes monopoly, property developers in singapore and poker. Will soon undertake a contiki trip that may include going to the Lower Valley of the Omo.

    My blog: http://www.primaboinca.com/view_profile.php?userid=5889534
  • 20 year-old Real Estate Agent Rusty from Saint-Paul, has hobbies and interests which includes monopoly, property developers in singapore and poker. Will soon undertake a contiki trip that may include going to the Lower Valley of the Omo.

    My blog: http://www.primaboinca.com/view_profile.php?userid=5889534
  • 20 year-old Real Estate Agent Rusty from Saint-Paul, has hobbies and interests which includes monopoly, property developers in singapore and poker. Will soon undertake a contiki trip that may include going to the Lower Valley of the Omo.

    My blog: http://www.primaboinca.com/view_profile.php?userid=5889534
  1. ↑ Dodge, Y. (2003) The Oxford Dictionary of Statistical Terms. OUP. ISBN 0-19-920613-9 (entry for admissible decision function)
  2. ↑ Template:Harvnb
  3. ↑ Template:Harvnb
Retrieved from "https://en.formulasearchengine.com/index.php?title=Equirectangular_projection&oldid=9272"