4
$\begingroup$

I am trying to create an example that applies fully parametric estimation. I am using a Gamma-Poisson distribution where the random variable is a Poisson random variable with mean $\lambda$ which has a Gamma distribution with parameters $\alpha$ and $\beta$. Also denoted as $X \sim \textrm{Gamma-Poisson}(\alpha,\beta)$ with probability mass function

\begin{equation*} f(x) = \frac{\Gamma{(x+\beta)}\alpha^{x}}{\Gamma(\beta)(1+\alpha)^{\beta+x}x!} \;\;\; x=0,1,2,... \end{equation*}

I am familiar with solving for MLE's but not entirely sure with this distribution. Currently what I have is below but I'm not sure about the $\Gamma$ function.

\begin{align*} L(\theta) &= \prod_{i=1}^{n} \frac{\Gamma{(x_i+\beta)}\alpha^{x_i}}{\Gamma(\beta)(1+\alpha)^{\beta+x_i}x_i!} \\ \textrm{ln} \; L(\theta) &= \sum_{i=1}^{n} \textrm{ln} \left(\frac{\Gamma{(x_i+\beta)}\alpha^{x_i}}{\Gamma(\beta)(1+\alpha)^{\beta+x_i}x_i!}\right) \\ &= \sum_{i=1}^{n} \big[\textrm{ln}\:\Gamma{(x_i+\beta)} + x_i\:\textrm{ln}\:\alpha - \textrm{ln}\:\Gamma(\beta) - (\beta+x_i)\:\textrm{ln}\:(1+\alpha) - \textrm{ln}\:(x_i!)\big] \\ & \; \vdots \\ \frac{\partial}{\partial\alpha}\;\textrm{ln}\;L(\theta) &= \dots = 0 \\ \hat{\alpha} &= \\ \frac{\partial}{\partial\beta}\;\textrm{ln}\;L(\theta) &= \dots = 0 \\ \hat{\beta} &= \end{align*}

$\endgroup$

2 Answers 2

1
$\begingroup$

You will not be able to solve this arithmetically, numerical analysis is required to find the MLE for beta distribution (and subsequently the distribution you are discussing here).

https://www.real-statistics.com/distribution-fitting/distribution-fitting-via-maximum-likelihood/fitting-beta-distribution-parameters-mle/

$\endgroup$
1
$\begingroup$

You can solve for 𝛼 in terms of 𝛽, and for 𝛽 you need to work with the Digamma Function.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.