Exponential Distribution - ML estimator of λ in τ parametrization












0














I am stuck with the following problem: We know that $x_1 cdots x_n$ ~ Ex$(1/lambda)$ and we are given the sum $sum_{i=1}^n x_i$. We want to find the maximum likelihood estimate of $lambda$, as well as its variance. Also we want to find a $(1-alpha)$ - confidence interval.



I know that E $ [x_i] = 1/lambda $ and that the likelihood is $L(frac{1}{lambda} | x_1cdots x_n) = lambda^n e^{-sum_i x_ilambda}$. Therefore the ML estimate should be
begin{align}
hat{lambda}= frac{n}{sum_i x_i}.
end{align}

However, I don't know how to calculate the expectation value or the variance of $hat{lambda}$ as this involes $E[frac{1}{sum_i x_i}]$ and I know that normally
begin{align}
E [frac{1}{X}] > frac{1}{E[X]}.
end{align}

Therefore I guess that E$[hat{lambda}] > lambda$, but I don't know how I would calculate it. Thank you already.










share|cite|improve this question






















  • $sum X_i sim Gamma(n,1/lambda)$, then find its variance.
    – John_Wick
    Nov 22 at 20:51












  • So var $[sum X_i ] = n/lambda^2$ ? I still don't see how this would help me as var $[1/X] neq 1/$var$[X]$, right?
    – F. Moe
    Nov 22 at 21:04










  • If $X$ is a random variable with support $Omega$ and pdf $f_X$, then $E(1/X)=int_Omega big(frac{1}{t}big)f_X(t)dt$, so like was mentioned use $Xsim Gamma(n,1/lambda)$
    – gd1035
    Nov 22 at 22:04


















0














I am stuck with the following problem: We know that $x_1 cdots x_n$ ~ Ex$(1/lambda)$ and we are given the sum $sum_{i=1}^n x_i$. We want to find the maximum likelihood estimate of $lambda$, as well as its variance. Also we want to find a $(1-alpha)$ - confidence interval.



I know that E $ [x_i] = 1/lambda $ and that the likelihood is $L(frac{1}{lambda} | x_1cdots x_n) = lambda^n e^{-sum_i x_ilambda}$. Therefore the ML estimate should be
begin{align}
hat{lambda}= frac{n}{sum_i x_i}.
end{align}

However, I don't know how to calculate the expectation value or the variance of $hat{lambda}$ as this involes $E[frac{1}{sum_i x_i}]$ and I know that normally
begin{align}
E [frac{1}{X}] > frac{1}{E[X]}.
end{align}

Therefore I guess that E$[hat{lambda}] > lambda$, but I don't know how I would calculate it. Thank you already.










share|cite|improve this question






















  • $sum X_i sim Gamma(n,1/lambda)$, then find its variance.
    – John_Wick
    Nov 22 at 20:51












  • So var $[sum X_i ] = n/lambda^2$ ? I still don't see how this would help me as var $[1/X] neq 1/$var$[X]$, right?
    – F. Moe
    Nov 22 at 21:04










  • If $X$ is a random variable with support $Omega$ and pdf $f_X$, then $E(1/X)=int_Omega big(frac{1}{t}big)f_X(t)dt$, so like was mentioned use $Xsim Gamma(n,1/lambda)$
    – gd1035
    Nov 22 at 22:04
















0












0








0







I am stuck with the following problem: We know that $x_1 cdots x_n$ ~ Ex$(1/lambda)$ and we are given the sum $sum_{i=1}^n x_i$. We want to find the maximum likelihood estimate of $lambda$, as well as its variance. Also we want to find a $(1-alpha)$ - confidence interval.



I know that E $ [x_i] = 1/lambda $ and that the likelihood is $L(frac{1}{lambda} | x_1cdots x_n) = lambda^n e^{-sum_i x_ilambda}$. Therefore the ML estimate should be
begin{align}
hat{lambda}= frac{n}{sum_i x_i}.
end{align}

However, I don't know how to calculate the expectation value or the variance of $hat{lambda}$ as this involes $E[frac{1}{sum_i x_i}]$ and I know that normally
begin{align}
E [frac{1}{X}] > frac{1}{E[X]}.
end{align}

Therefore I guess that E$[hat{lambda}] > lambda$, but I don't know how I would calculate it. Thank you already.










share|cite|improve this question













I am stuck with the following problem: We know that $x_1 cdots x_n$ ~ Ex$(1/lambda)$ and we are given the sum $sum_{i=1}^n x_i$. We want to find the maximum likelihood estimate of $lambda$, as well as its variance. Also we want to find a $(1-alpha)$ - confidence interval.



I know that E $ [x_i] = 1/lambda $ and that the likelihood is $L(frac{1}{lambda} | x_1cdots x_n) = lambda^n e^{-sum_i x_ilambda}$. Therefore the ML estimate should be
begin{align}
hat{lambda}= frac{n}{sum_i x_i}.
end{align}

However, I don't know how to calculate the expectation value or the variance of $hat{lambda}$ as this involes $E[frac{1}{sum_i x_i}]$ and I know that normally
begin{align}
E [frac{1}{X}] > frac{1}{E[X]}.
end{align}

Therefore I guess that E$[hat{lambda}] > lambda$, but I don't know how I would calculate it. Thank you already.







statistics exponential-distribution






share|cite|improve this question













share|cite|improve this question











share|cite|improve this question




share|cite|improve this question










asked Nov 22 at 20:47









F. Moe

205




205












  • $sum X_i sim Gamma(n,1/lambda)$, then find its variance.
    – John_Wick
    Nov 22 at 20:51












  • So var $[sum X_i ] = n/lambda^2$ ? I still don't see how this would help me as var $[1/X] neq 1/$var$[X]$, right?
    – F. Moe
    Nov 22 at 21:04










  • If $X$ is a random variable with support $Omega$ and pdf $f_X$, then $E(1/X)=int_Omega big(frac{1}{t}big)f_X(t)dt$, so like was mentioned use $Xsim Gamma(n,1/lambda)$
    – gd1035
    Nov 22 at 22:04




















  • $sum X_i sim Gamma(n,1/lambda)$, then find its variance.
    – John_Wick
    Nov 22 at 20:51












  • So var $[sum X_i ] = n/lambda^2$ ? I still don't see how this would help me as var $[1/X] neq 1/$var$[X]$, right?
    – F. Moe
    Nov 22 at 21:04










  • If $X$ is a random variable with support $Omega$ and pdf $f_X$, then $E(1/X)=int_Omega big(frac{1}{t}big)f_X(t)dt$, so like was mentioned use $Xsim Gamma(n,1/lambda)$
    – gd1035
    Nov 22 at 22:04


















$sum X_i sim Gamma(n,1/lambda)$, then find its variance.
– John_Wick
Nov 22 at 20:51






$sum X_i sim Gamma(n,1/lambda)$, then find its variance.
– John_Wick
Nov 22 at 20:51














So var $[sum X_i ] = n/lambda^2$ ? I still don't see how this would help me as var $[1/X] neq 1/$var$[X]$, right?
– F. Moe
Nov 22 at 21:04




So var $[sum X_i ] = n/lambda^2$ ? I still don't see how this would help me as var $[1/X] neq 1/$var$[X]$, right?
– F. Moe
Nov 22 at 21:04












If $X$ is a random variable with support $Omega$ and pdf $f_X$, then $E(1/X)=int_Omega big(frac{1}{t}big)f_X(t)dt$, so like was mentioned use $Xsim Gamma(n,1/lambda)$
– gd1035
Nov 22 at 22:04






If $X$ is a random variable with support $Omega$ and pdf $f_X$, then $E(1/X)=int_Omega big(frac{1}{t}big)f_X(t)dt$, so like was mentioned use $Xsim Gamma(n,1/lambda)$
– gd1035
Nov 22 at 22:04












1 Answer
1






active

oldest

votes


















1














$S=sum X_i sim Gamma(n, 1/lambda).$ If $Ssim Gamma(n,a)$ then $E(S^t)=int_0^infty x^t frac{1}{Gamma(n)a^n}e^{-x/a}x^{n-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}int_0^infty frac{1}{Gamma(n+t)a^{n+t}}e^{-x/a}x^{n+t-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}$ for all $t>-n$.
$Var(n/S)=E(n^2/S^2)-E^2(n/S)=n^2left(frac{Gamma(n-2)a^{n-2}}{Gamma(n)a^n}-left(frac{Gamma(n-1)a^{n-1}}{Gamma(n)a^n}right)^2right)=n^2left(frac{1}{(n-1)(n-2)a^2}-left(frac{1}{(n-1)a}right)^2right)=frac{n^2}{a^2(n-1)^2(n-2)}.$






share|cite|improve this answer























  • Thank you for clarifying! In your calculation of $E(S^t)$ you forgot the $a$ in the exponent, right? So it would be $e^{-x/a}$. Am I right that $E(hat{lambda})$ then is $E(hat{lambda}) = n frac{Gamma(n-1)}{Gamma(n)} lambda$?
    – F. Moe
    Nov 22 at 22:53












  • Yes, that is correct.
    – John_Wick
    Nov 22 at 22:55










  • Great! Just one more question: will the confidence interval for $lambda$ be just taken from the regular Gamma quantiles, i.e. from Gamma$(1,frac{1}{lambda})$ as Ex$(frac{1}{lambda}) = $Gamma$(1,frac{1}{lambda})$ ?
    – F. Moe
    Nov 22 at 23:08












  • Yes, you can find the bounds for CI using quantiles for Gamma.
    – John_Wick
    Nov 22 at 23:10










  • Thanks again. You really helped me clarify all this in my mind!
    – F. Moe
    Nov 22 at 23:12











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
});


}
});














draft saved

draft discarded


















StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3009639%2fexponential-distribution-ml-estimator-of-%25ce%25bb-in-%25cf%2584-parametrization%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









1














$S=sum X_i sim Gamma(n, 1/lambda).$ If $Ssim Gamma(n,a)$ then $E(S^t)=int_0^infty x^t frac{1}{Gamma(n)a^n}e^{-x/a}x^{n-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}int_0^infty frac{1}{Gamma(n+t)a^{n+t}}e^{-x/a}x^{n+t-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}$ for all $t>-n$.
$Var(n/S)=E(n^2/S^2)-E^2(n/S)=n^2left(frac{Gamma(n-2)a^{n-2}}{Gamma(n)a^n}-left(frac{Gamma(n-1)a^{n-1}}{Gamma(n)a^n}right)^2right)=n^2left(frac{1}{(n-1)(n-2)a^2}-left(frac{1}{(n-1)a}right)^2right)=frac{n^2}{a^2(n-1)^2(n-2)}.$






share|cite|improve this answer























  • Thank you for clarifying! In your calculation of $E(S^t)$ you forgot the $a$ in the exponent, right? So it would be $e^{-x/a}$. Am I right that $E(hat{lambda})$ then is $E(hat{lambda}) = n frac{Gamma(n-1)}{Gamma(n)} lambda$?
    – F. Moe
    Nov 22 at 22:53












  • Yes, that is correct.
    – John_Wick
    Nov 22 at 22:55










  • Great! Just one more question: will the confidence interval for $lambda$ be just taken from the regular Gamma quantiles, i.e. from Gamma$(1,frac{1}{lambda})$ as Ex$(frac{1}{lambda}) = $Gamma$(1,frac{1}{lambda})$ ?
    – F. Moe
    Nov 22 at 23:08












  • Yes, you can find the bounds for CI using quantiles for Gamma.
    – John_Wick
    Nov 22 at 23:10










  • Thanks again. You really helped me clarify all this in my mind!
    – F. Moe
    Nov 22 at 23:12
















1














$S=sum X_i sim Gamma(n, 1/lambda).$ If $Ssim Gamma(n,a)$ then $E(S^t)=int_0^infty x^t frac{1}{Gamma(n)a^n}e^{-x/a}x^{n-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}int_0^infty frac{1}{Gamma(n+t)a^{n+t}}e^{-x/a}x^{n+t-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}$ for all $t>-n$.
$Var(n/S)=E(n^2/S^2)-E^2(n/S)=n^2left(frac{Gamma(n-2)a^{n-2}}{Gamma(n)a^n}-left(frac{Gamma(n-1)a^{n-1}}{Gamma(n)a^n}right)^2right)=n^2left(frac{1}{(n-1)(n-2)a^2}-left(frac{1}{(n-1)a}right)^2right)=frac{n^2}{a^2(n-1)^2(n-2)}.$






share|cite|improve this answer























  • Thank you for clarifying! In your calculation of $E(S^t)$ you forgot the $a$ in the exponent, right? So it would be $e^{-x/a}$. Am I right that $E(hat{lambda})$ then is $E(hat{lambda}) = n frac{Gamma(n-1)}{Gamma(n)} lambda$?
    – F. Moe
    Nov 22 at 22:53












  • Yes, that is correct.
    – John_Wick
    Nov 22 at 22:55










  • Great! Just one more question: will the confidence interval for $lambda$ be just taken from the regular Gamma quantiles, i.e. from Gamma$(1,frac{1}{lambda})$ as Ex$(frac{1}{lambda}) = $Gamma$(1,frac{1}{lambda})$ ?
    – F. Moe
    Nov 22 at 23:08












  • Yes, you can find the bounds for CI using quantiles for Gamma.
    – John_Wick
    Nov 22 at 23:10










  • Thanks again. You really helped me clarify all this in my mind!
    – F. Moe
    Nov 22 at 23:12














1












1








1






$S=sum X_i sim Gamma(n, 1/lambda).$ If $Ssim Gamma(n,a)$ then $E(S^t)=int_0^infty x^t frac{1}{Gamma(n)a^n}e^{-x/a}x^{n-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}int_0^infty frac{1}{Gamma(n+t)a^{n+t}}e^{-x/a}x^{n+t-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}$ for all $t>-n$.
$Var(n/S)=E(n^2/S^2)-E^2(n/S)=n^2left(frac{Gamma(n-2)a^{n-2}}{Gamma(n)a^n}-left(frac{Gamma(n-1)a^{n-1}}{Gamma(n)a^n}right)^2right)=n^2left(frac{1}{(n-1)(n-2)a^2}-left(frac{1}{(n-1)a}right)^2right)=frac{n^2}{a^2(n-1)^2(n-2)}.$






share|cite|improve this answer














$S=sum X_i sim Gamma(n, 1/lambda).$ If $Ssim Gamma(n,a)$ then $E(S^t)=int_0^infty x^t frac{1}{Gamma(n)a^n}e^{-x/a}x^{n-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}int_0^infty frac{1}{Gamma(n+t)a^{n+t}}e^{-x/a}x^{n+t-1}dx=frac{Gamma(n+t)a^{n+t}}{Gamma(n)a^n}$ for all $t>-n$.
$Var(n/S)=E(n^2/S^2)-E^2(n/S)=n^2left(frac{Gamma(n-2)a^{n-2}}{Gamma(n)a^n}-left(frac{Gamma(n-1)a^{n-1}}{Gamma(n)a^n}right)^2right)=n^2left(frac{1}{(n-1)(n-2)a^2}-left(frac{1}{(n-1)a}right)^2right)=frac{n^2}{a^2(n-1)^2(n-2)}.$







share|cite|improve this answer














share|cite|improve this answer



share|cite|improve this answer








edited Nov 22 at 22:56

























answered Nov 22 at 21:19









John_Wick

1,346111




1,346111












  • Thank you for clarifying! In your calculation of $E(S^t)$ you forgot the $a$ in the exponent, right? So it would be $e^{-x/a}$. Am I right that $E(hat{lambda})$ then is $E(hat{lambda}) = n frac{Gamma(n-1)}{Gamma(n)} lambda$?
    – F. Moe
    Nov 22 at 22:53












  • Yes, that is correct.
    – John_Wick
    Nov 22 at 22:55










  • Great! Just one more question: will the confidence interval for $lambda$ be just taken from the regular Gamma quantiles, i.e. from Gamma$(1,frac{1}{lambda})$ as Ex$(frac{1}{lambda}) = $Gamma$(1,frac{1}{lambda})$ ?
    – F. Moe
    Nov 22 at 23:08












  • Yes, you can find the bounds for CI using quantiles for Gamma.
    – John_Wick
    Nov 22 at 23:10










  • Thanks again. You really helped me clarify all this in my mind!
    – F. Moe
    Nov 22 at 23:12


















  • Thank you for clarifying! In your calculation of $E(S^t)$ you forgot the $a$ in the exponent, right? So it would be $e^{-x/a}$. Am I right that $E(hat{lambda})$ then is $E(hat{lambda}) = n frac{Gamma(n-1)}{Gamma(n)} lambda$?
    – F. Moe
    Nov 22 at 22:53












  • Yes, that is correct.
    – John_Wick
    Nov 22 at 22:55










  • Great! Just one more question: will the confidence interval for $lambda$ be just taken from the regular Gamma quantiles, i.e. from Gamma$(1,frac{1}{lambda})$ as Ex$(frac{1}{lambda}) = $Gamma$(1,frac{1}{lambda})$ ?
    – F. Moe
    Nov 22 at 23:08












  • Yes, you can find the bounds for CI using quantiles for Gamma.
    – John_Wick
    Nov 22 at 23:10










  • Thanks again. You really helped me clarify all this in my mind!
    – F. Moe
    Nov 22 at 23:12
















Thank you for clarifying! In your calculation of $E(S^t)$ you forgot the $a$ in the exponent, right? So it would be $e^{-x/a}$. Am I right that $E(hat{lambda})$ then is $E(hat{lambda}) = n frac{Gamma(n-1)}{Gamma(n)} lambda$?
– F. Moe
Nov 22 at 22:53






Thank you for clarifying! In your calculation of $E(S^t)$ you forgot the $a$ in the exponent, right? So it would be $e^{-x/a}$. Am I right that $E(hat{lambda})$ then is $E(hat{lambda}) = n frac{Gamma(n-1)}{Gamma(n)} lambda$?
– F. Moe
Nov 22 at 22:53














Yes, that is correct.
– John_Wick
Nov 22 at 22:55




Yes, that is correct.
– John_Wick
Nov 22 at 22:55












Great! Just one more question: will the confidence interval for $lambda$ be just taken from the regular Gamma quantiles, i.e. from Gamma$(1,frac{1}{lambda})$ as Ex$(frac{1}{lambda}) = $Gamma$(1,frac{1}{lambda})$ ?
– F. Moe
Nov 22 at 23:08






Great! Just one more question: will the confidence interval for $lambda$ be just taken from the regular Gamma quantiles, i.e. from Gamma$(1,frac{1}{lambda})$ as Ex$(frac{1}{lambda}) = $Gamma$(1,frac{1}{lambda})$ ?
– F. Moe
Nov 22 at 23:08














Yes, you can find the bounds for CI using quantiles for Gamma.
– John_Wick
Nov 22 at 23:10




Yes, you can find the bounds for CI using quantiles for Gamma.
– John_Wick
Nov 22 at 23:10












Thanks again. You really helped me clarify all this in my mind!
– F. Moe
Nov 22 at 23:12




Thanks again. You really helped me clarify all this in my mind!
– F. Moe
Nov 22 at 23:12


















draft saved

draft discarded




















































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.





Some of your past answers have not been well-received, and you're in danger of being blocked from answering.


Please pay close attention to the following guidance:


  • 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.


To learn more, see our tips on writing great answers.




draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3009639%2fexponential-distribution-ml-estimator-of-%25ce%25bb-in-%25cf%2584-parametrization%23new-answer', 'question_page');
}
);

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







Popular posts from this blog

Plaza Victoria

Puebla de Zaragoza

Musa