# Approximating bias for the MLE of an exponential distribution PDF

by llamaro25   Last Updated September 12, 2019 03:19 AM

The exponential distribution is a probability distribution often used to model waiting or survival times. The version that we will look at has a probability density function of the form

$$p(y | v) = exp((-e^-v)y - v)$$

where $$0, i.e., y can take on the values of non-negative real numbers. In this form it has one parameters: a log-scale parameter $$v$$. If a random variable follows a gamma distribution with log-scale $$v$$ we say that $$Y ~Exp(v)$$. If Y ~Exp(v), then $$E[Y]= e^v$$ and V[Y]= e^(2v).

The question is;

Determine the approximate bias of the maximum likelihood estimator $$\hat{v}$$ of $$v$$ for the exponential distribution.

The MLE that I have gotten is $$\hat{v}= \ln (m/n)$$ where $$m=\sum_1^n y_i$$

I know that the formula for bias is $$b(\hat{v}) = E(\hat{v}) - v$$

Here's my attempt to calculate the bias.

$$b(\hat{v}) = ln(m/n) - v$$ since $$m/n$$ is the sample mean, $$m/n = E[Y]= E(m/n) = e^v$$

This implies that

$$b(\hat{v}) = ln(e^v)- v = 0$$ So the bias is 0? I'm not sure if I calculated the MLE and bias correctly as I'm new to statistics

Tags :