Revision as of 10:13, 20 May 2013 by Rhea (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)


Examples of Parameter Estimation based on Maximum Likelihood (MLE): the exponential distribution and the geometric distribution

for ECE662: Decision Theory

Complement to Lecture 7: "Comparison of Maximum likelihood (MLE) and Bayesian Parameter Estimation"


Exponential Distribution

Let $ {X}_{1}, {X}_{2}, {X}_{3}.....{X}_{n} $ be a random sample from the exponential distribution with p.d.f.

$ f(x;\theta)=\frac{1}{\theta}{e}^{\frac{-x}{\theta}} 0<x<\infty, \theta\in\Omega=\{\theta|0<\theta<\infty\} $

The likelihood function is given by:

$ L(\theta)=L\left(\theta;{x}_{1},{x}_{2}...{x}_{n} \right)=\left(\frac{1}{\theta}{e}^{\frac{{-x}_{1}}{\theta}}\right)\left(\frac{1}{\theta}{e}^{\frac{{-x}_{2}}{\theta}}\right)...\left(\frac{1}{\theta}{e}^{\frac{{-x}_{n}}{\theta}} \right)=\frac{1}{{\theta}^{n}}exp\left(\frac{-\sum_{1}^{n}{x}_{i}}{\theta} \right) $

Taking log, we get,

$ lnL\left(\theta\right)=-\left(n \right)ln\left(\theta\right) -\frac{1}{\theta}\sum_{1}^{n}{x}_{i}, 0<\theta<\infty $

Differentiating the above expression, and equating to zero, we get

$ \frac{d\left[lnL\left(\theta\right) \right]}{d\theta}=\frac{-\left(n \right)}{\left(\theta\right)} +\frac{1}{{\theta}^{2}}\sum_{1}^{n}{x}_{i}=0 $

The solution of equation for $ \theta $ is:

$ \theta=\frac{\sum_{1}^{n}{x}_{i}}{n} $

Thus, the maximum likelihood estimator of $ \Theta $ is

$ \Theta=\frac{\sum_{1}^{n}{X}_{i}}{n} $


Geometric Distribution

Let $ {X}_{1}, {X}_{2}, {X}_{3}.....{X}_{n} $ be a random sample from the geometric distribution with p.d.f.

$ f\left(x;p \right)={\left(1-p \right)}^{x-1}p, x=1,2,3.... $

The likelihood function is given by:

$ L\left(p \right)={\left(1-p \right)}^{{x}_{1}-1}p {\left(1-p \right)}^{{x}_{2}-1}p...{\left(1-p \right)}^{{x}_{n}-1}p ={p}^{n}{\left(1-p \right)}^{\sum_{1}^{n}{x}_{i}-n} $

Taking log,

$ lnL\left(p \right)= nln{p}+\left(\sum_{1}^{n}{x}_{i}-n \right)ln{\left(1-p \right)} $

Differentiating and equating to zero, we get,

$ \frac{d\left[lnL\left(p \right)\right]}{dp}=\frac{n}{p} -\frac{\left(\sum_{1}^{n}{x}_{i}-n \right)}{\left(1-p \right)}=0 $

Therefore,

$ p=\frac{n}{\left(\sum_{1}^{n}{x}_{i} \right)} $

So, the maximum likelihood estimator of P is:

$ P=\frac{n}{\left(\sum_{1}^{n}{X}_{i} \right)}=\frac{1}{X} $

This agrees with the intuition because, in n observations of a geometric random variable, there are n successes in the $ \sum_{1}^{n}{X}_{i} $ trials. Thus the estimate of p is the number of successes divided by the total number of trials.


More examples: Binomial and Poisson Distributions

Back to Lecture 7: "Comparison of Maximum likelihood (MLE) and Bayesian Parameter Estimation"

Back to ECE662, Spring 2008, Prof. Boutin

Alumni Liaison

Correspondence Chess Grandmaster and Purdue Alumni

Prof. Dan Fleetwood