Hi - I encounter two problems with
SystemFit. I have a matrix of 20 variables (380000 observations each).
I am trying to fit using "2SLS" in Systemfit because I want to be able
to simulate new observations with the resulting model. The first
problem I found is that it ran out of memory given that I have 2GB of
RAM. How can I circumvent this? Can I do some mixing or bootstrapping
with small sample each time instead?Another problem is I got
system is computationally singular when I reduce the size of
observation to 100000 instead of 380000. Any help would be appreciated.
I think my variables are not really well correlated to that extent. adschai
[[alternative HTML version deleted]]
