GLM matrix

IAI Nov 2013 q12



Would you please explain how the avg loss to layer is estimated in (iV) ??

I'm really sorry! I'm unable to to edit the title of the thread! its neither abt matrices nor glms! I meant to post a question on glms, but I worked it out myself but forgot to change the title!

 

Where is the question?

Based on what you have said i would go about it as follows:

The parameter vector Beta has 6 rows
1 for the intercept
1 for the car type (2-1 as one of them become the base line)
2 for age (3-1 as one of them becomes the base line)
2 for the interaction term (6 possible combination 4 of which contain the baseline variable either singularly, car type or age, or both together)

This would translate to a matrix with 6 columns. The number of rows would depend on the number of observations.

Btw.... not sure about the number of parameters for the interaction term. Maybe someone else can shed some light on t his?

 

Im really sorry, I did manage to do the glm question. It was in the IAI Nov 2013 paper.