Sample median of Cauchy distribution is consistent. How?
$begingroup$
When we use chebyshev's inequality to show whether an estimator is consistent or not, we require the mean square error of the estimator and I do not know sample median's probability distribution. So please advice how this can be shown.
statistics statistical-inference quantile
$endgroup$
add a comment |
$begingroup$
When we use chebyshev's inequality to show whether an estimator is consistent or not, we require the mean square error of the estimator and I do not know sample median's probability distribution. So please advice how this can be shown.
statistics statistical-inference quantile
$endgroup$
1
$begingroup$
More generally I think you can say that all empirical quantiles converge to the corresponding true quantiles for any continuous distribution as the sample size tends to infinity.
$endgroup$
– dsaxton
Jun 26 '15 at 3:26
$begingroup$
For every $x$, the median $M_n$ of an i.i.d. sample of size $2n+1$ is such that $P(M_nleqslant x)=P(Y_n^xgeqslant n)$, where $Y_n^x$ is binomial $(2n+1,p(x))$ and $p(x)=P(Xleqslant x)$. If $x<0$, by the law of large numbers, $Y_n^x/(2n+1)to p(x)$ and $p(x)<frac12$ hence $P(Y_n^xgeqslant n)to0$. Likewise, if $x>0$, $P(Y_n^xgeqslant n)to1$. Thus, $M_nto0$ in distribution (and in probability).
$endgroup$
– Did
Jun 26 '15 at 6:07
add a comment |
$begingroup$
When we use chebyshev's inequality to show whether an estimator is consistent or not, we require the mean square error of the estimator and I do not know sample median's probability distribution. So please advice how this can be shown.
statistics statistical-inference quantile
$endgroup$
When we use chebyshev's inequality to show whether an estimator is consistent or not, we require the mean square error of the estimator and I do not know sample median's probability distribution. So please advice how this can be shown.
statistics statistical-inference quantile
statistics statistical-inference quantile
edited Jun 26 '15 at 5:54
A.D
3,9281235
3,9281235
asked Jun 25 '15 at 13:25
Kiran PrajapatKiran Prajapat
135
135
1
$begingroup$
More generally I think you can say that all empirical quantiles converge to the corresponding true quantiles for any continuous distribution as the sample size tends to infinity.
$endgroup$
– dsaxton
Jun 26 '15 at 3:26
$begingroup$
For every $x$, the median $M_n$ of an i.i.d. sample of size $2n+1$ is such that $P(M_nleqslant x)=P(Y_n^xgeqslant n)$, where $Y_n^x$ is binomial $(2n+1,p(x))$ and $p(x)=P(Xleqslant x)$. If $x<0$, by the law of large numbers, $Y_n^x/(2n+1)to p(x)$ and $p(x)<frac12$ hence $P(Y_n^xgeqslant n)to0$. Likewise, if $x>0$, $P(Y_n^xgeqslant n)to1$. Thus, $M_nto0$ in distribution (and in probability).
$endgroup$
– Did
Jun 26 '15 at 6:07
add a comment |
1
$begingroup$
More generally I think you can say that all empirical quantiles converge to the corresponding true quantiles for any continuous distribution as the sample size tends to infinity.
$endgroup$
– dsaxton
Jun 26 '15 at 3:26
$begingroup$
For every $x$, the median $M_n$ of an i.i.d. sample of size $2n+1$ is such that $P(M_nleqslant x)=P(Y_n^xgeqslant n)$, where $Y_n^x$ is binomial $(2n+1,p(x))$ and $p(x)=P(Xleqslant x)$. If $x<0$, by the law of large numbers, $Y_n^x/(2n+1)to p(x)$ and $p(x)<frac12$ hence $P(Y_n^xgeqslant n)to0$. Likewise, if $x>0$, $P(Y_n^xgeqslant n)to1$. Thus, $M_nto0$ in distribution (and in probability).
$endgroup$
– Did
Jun 26 '15 at 6:07
1
1
$begingroup$
More generally I think you can say that all empirical quantiles converge to the corresponding true quantiles for any continuous distribution as the sample size tends to infinity.
$endgroup$
– dsaxton
Jun 26 '15 at 3:26
$begingroup$
More generally I think you can say that all empirical quantiles converge to the corresponding true quantiles for any continuous distribution as the sample size tends to infinity.
$endgroup$
– dsaxton
Jun 26 '15 at 3:26
$begingroup$
For every $x$, the median $M_n$ of an i.i.d. sample of size $2n+1$ is such that $P(M_nleqslant x)=P(Y_n^xgeqslant n)$, where $Y_n^x$ is binomial $(2n+1,p(x))$ and $p(x)=P(Xleqslant x)$. If $x<0$, by the law of large numbers, $Y_n^x/(2n+1)to p(x)$ and $p(x)<frac12$ hence $P(Y_n^xgeqslant n)to0$. Likewise, if $x>0$, $P(Y_n^xgeqslant n)to1$. Thus, $M_nto0$ in distribution (and in probability).
$endgroup$
– Did
Jun 26 '15 at 6:07
$begingroup$
For every $x$, the median $M_n$ of an i.i.d. sample of size $2n+1$ is such that $P(M_nleqslant x)=P(Y_n^xgeqslant n)$, where $Y_n^x$ is binomial $(2n+1,p(x))$ and $p(x)=P(Xleqslant x)$. If $x<0$, by the law of large numbers, $Y_n^x/(2n+1)to p(x)$ and $p(x)<frac12$ hence $P(Y_n^xgeqslant n)to0$. Likewise, if $x>0$, $P(Y_n^xgeqslant n)to1$. Thus, $M_nto0$ in distribution (and in probability).
$endgroup$
– Did
Jun 26 '15 at 6:07
add a comment |
1 Answer
1
active
oldest
votes
$begingroup$
Partial answer
For uneven samples $n=2k+1$ the distribution of the median of a distribution with probability density function $f(x)$ and cumulative density function F(x) can be expressed as:
$$f_{text{median n= 2k+1}}(x) = frac{(2k+1)!}{k! k!} F(x)^kf(x) (1-F(x))^k$$
which relates to the sampling of a number at $x$ (the middle term $f(x)$), having $k$ samples that are below $x$ (the first term $F(x)^k$), having $k$ samples above $x$ (the last term $F(x)^k$), and the number of ways that these numbers can be arranged (the factor at the beginning $(2k+1)!/(k!k!)$).
For the Cauchy distribution this becomes:
$$f_{text{Cauchy median $n=2k+1$}}(x) = frac{(2k+1)!}{k! k!} left( frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] } right) times left( 0.25 - frac{text{tan}^{-1}left(frac{x-x_0}{gamma} right)}{pi^2} vphantom{frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] }}right)^k $$
The variance does seem to be finite for $k geq 2$ (at least checking in wolfram alpha). I can imagine it might be possible to obtain an explicit expression. It seems like the indefinite integral can be expressed in terms of inverse tangent, logarithm and polylogarithm functions. But it is not a very pretty looking expression.
$endgroup$
add a comment |
Your Answer
StackExchange.ifUsing("editor", function () {
return StackExchange.using("mathjaxEditing", function () {
StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix) {
StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
});
});
}, "mathjax-editing");
StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "69"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);
StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});
function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
noCode: true, onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});
}
});
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f1338845%2fsample-median-of-cauchy-distribution-is-consistent-how%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
1 Answer
1
active
oldest
votes
1 Answer
1
active
oldest
votes
active
oldest
votes
active
oldest
votes
$begingroup$
Partial answer
For uneven samples $n=2k+1$ the distribution of the median of a distribution with probability density function $f(x)$ and cumulative density function F(x) can be expressed as:
$$f_{text{median n= 2k+1}}(x) = frac{(2k+1)!}{k! k!} F(x)^kf(x) (1-F(x))^k$$
which relates to the sampling of a number at $x$ (the middle term $f(x)$), having $k$ samples that are below $x$ (the first term $F(x)^k$), having $k$ samples above $x$ (the last term $F(x)^k$), and the number of ways that these numbers can be arranged (the factor at the beginning $(2k+1)!/(k!k!)$).
For the Cauchy distribution this becomes:
$$f_{text{Cauchy median $n=2k+1$}}(x) = frac{(2k+1)!}{k! k!} left( frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] } right) times left( 0.25 - frac{text{tan}^{-1}left(frac{x-x_0}{gamma} right)}{pi^2} vphantom{frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] }}right)^k $$
The variance does seem to be finite for $k geq 2$ (at least checking in wolfram alpha). I can imagine it might be possible to obtain an explicit expression. It seems like the indefinite integral can be expressed in terms of inverse tangent, logarithm and polylogarithm functions. But it is not a very pretty looking expression.
$endgroup$
add a comment |
$begingroup$
Partial answer
For uneven samples $n=2k+1$ the distribution of the median of a distribution with probability density function $f(x)$ and cumulative density function F(x) can be expressed as:
$$f_{text{median n= 2k+1}}(x) = frac{(2k+1)!}{k! k!} F(x)^kf(x) (1-F(x))^k$$
which relates to the sampling of a number at $x$ (the middle term $f(x)$), having $k$ samples that are below $x$ (the first term $F(x)^k$), having $k$ samples above $x$ (the last term $F(x)^k$), and the number of ways that these numbers can be arranged (the factor at the beginning $(2k+1)!/(k!k!)$).
For the Cauchy distribution this becomes:
$$f_{text{Cauchy median $n=2k+1$}}(x) = frac{(2k+1)!}{k! k!} left( frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] } right) times left( 0.25 - frac{text{tan}^{-1}left(frac{x-x_0}{gamma} right)}{pi^2} vphantom{frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] }}right)^k $$
The variance does seem to be finite for $k geq 2$ (at least checking in wolfram alpha). I can imagine it might be possible to obtain an explicit expression. It seems like the indefinite integral can be expressed in terms of inverse tangent, logarithm and polylogarithm functions. But it is not a very pretty looking expression.
$endgroup$
add a comment |
$begingroup$
Partial answer
For uneven samples $n=2k+1$ the distribution of the median of a distribution with probability density function $f(x)$ and cumulative density function F(x) can be expressed as:
$$f_{text{median n= 2k+1}}(x) = frac{(2k+1)!}{k! k!} F(x)^kf(x) (1-F(x))^k$$
which relates to the sampling of a number at $x$ (the middle term $f(x)$), having $k$ samples that are below $x$ (the first term $F(x)^k$), having $k$ samples above $x$ (the last term $F(x)^k$), and the number of ways that these numbers can be arranged (the factor at the beginning $(2k+1)!/(k!k!)$).
For the Cauchy distribution this becomes:
$$f_{text{Cauchy median $n=2k+1$}}(x) = frac{(2k+1)!}{k! k!} left( frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] } right) times left( 0.25 - frac{text{tan}^{-1}left(frac{x-x_0}{gamma} right)}{pi^2} vphantom{frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] }}right)^k $$
The variance does seem to be finite for $k geq 2$ (at least checking in wolfram alpha). I can imagine it might be possible to obtain an explicit expression. It seems like the indefinite integral can be expressed in terms of inverse tangent, logarithm and polylogarithm functions. But it is not a very pretty looking expression.
$endgroup$
Partial answer
For uneven samples $n=2k+1$ the distribution of the median of a distribution with probability density function $f(x)$ and cumulative density function F(x) can be expressed as:
$$f_{text{median n= 2k+1}}(x) = frac{(2k+1)!}{k! k!} F(x)^kf(x) (1-F(x))^k$$
which relates to the sampling of a number at $x$ (the middle term $f(x)$), having $k$ samples that are below $x$ (the first term $F(x)^k$), having $k$ samples above $x$ (the last term $F(x)^k$), and the number of ways that these numbers can be arranged (the factor at the beginning $(2k+1)!/(k!k!)$).
For the Cauchy distribution this becomes:
$$f_{text{Cauchy median $n=2k+1$}}(x) = frac{(2k+1)!}{k! k!} left( frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] } right) times left( 0.25 - frac{text{tan}^{-1}left(frac{x-x_0}{gamma} right)}{pi^2} vphantom{frac{1}{pi gamma left[1 + left(frac{x-x_0}{gamma} right)^2 right] }}right)^k $$
The variance does seem to be finite for $k geq 2$ (at least checking in wolfram alpha). I can imagine it might be possible to obtain an explicit expression. It seems like the indefinite integral can be expressed in terms of inverse tangent, logarithm and polylogarithm functions. But it is not a very pretty looking expression.
edited Jan 12 at 17:51
answered Jan 12 at 17:46
Martijn WeteringsMartijn Weterings
17010
17010
add a comment |
add a comment |
Thanks for contributing an answer to Mathematics Stack Exchange!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
Use MathJax to format equations. MathJax reference.
To learn more, see our tips on writing great answers.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f1338845%2fsample-median-of-cauchy-distribution-is-consistent-how%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
1
$begingroup$
More generally I think you can say that all empirical quantiles converge to the corresponding true quantiles for any continuous distribution as the sample size tends to infinity.
$endgroup$
– dsaxton
Jun 26 '15 at 3:26
$begingroup$
For every $x$, the median $M_n$ of an i.i.d. sample of size $2n+1$ is such that $P(M_nleqslant x)=P(Y_n^xgeqslant n)$, where $Y_n^x$ is binomial $(2n+1,p(x))$ and $p(x)=P(Xleqslant x)$. If $x<0$, by the law of large numbers, $Y_n^x/(2n+1)to p(x)$ and $p(x)<frac12$ hence $P(Y_n^xgeqslant n)to0$. Likewise, if $x>0$, $P(Y_n^xgeqslant n)to1$. Thus, $M_nto0$ in distribution (and in probability).
$endgroup$
– Did
Jun 26 '15 at 6:07