"Konrad Den Ende" <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>...
> The function
> L = \sum_{i=1}^b \sum_{j=1}^k (y_ij - \beta_i - \mu_j)^2
> is given. The task is to differentiate it with respect to all b times k
> parameters (which isn't very difficult) and use it to calculate the
> estimates for all the \beta_i's and \mu_j's.Either you're making a mistake with calculating the derivative OR you're making a mistake with using that to find the least squares (or equivalently, the maximum of the likelihood at the normal) solution. To narrow it down, could you say what you think the derivative of the above loss function is with respect to beta_i? (since it's symmetric in all of the parameters - with possible interchange of b and k - you only need to give one; if you can get one right, all the other betas and mus work the same) (This is homework, right?) Glen . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
