# MLE of bivariate normal

• November 26th 2010, 11:16 PM
Sambit
MLE of bivariate normal
how to calculate the MLEs of the 5 parameters of a bivariate normal population, the parameters being $\mu_x,\mu_y,\sigma_x,\sigma_y,\rho$?

while trying to solve for $\mu_x$, i got the following equation
$\sum{\frac{(x-\mu_x)}{\sigma_x^2}} = \rho\sum{\frac{(y-\mu_y)}{\sigma_x\sigma_y}}$

is this correct? how can i solve for $\mu_x$ from here?
• November 26th 2010, 11:55 PM
matheagle
you need to take the partial derivative wrt to all 5 parameters
It's smart to take the logarithm of the joint density first
• November 27th 2010, 05:14 AM
Sambit
I took the logarithm and differentiate partially wrt $\mu$ and got that
• November 27th 2010, 08:11 AM
matheagle
you need to solve all five together, the MLE of $\mu_X$ should be $\bar X$
and those for the population variances should be the sample variances with n in the denomiator.