Displaying 1 result from an estimated 1 matches for "ssr2".
Did you mean:
sse2
2010 May 17
0
plm(..., model="within", effect="twoways") is very slow on unablanaced data (was: Re: Regressions with fixed-effect in R)
...ot;twoways")
I would have went ahead and computed the within R-sq manually, from
pd_fe2, however the below computes the "individual" within R-sq
RSS2 <- sum(residuals(pd_fe2)^2); RSS2
TSS2 <- plm:::tss.plm(pd_fe2); TSS2
1 - RSS2/TSS2
#[1] 0.52888
while the below yields zero
SSR2 <- sum(fitted(pd_fe2)^2); SSR2
SSR2/TSS2
#[1] 0
because
fitted(pd_fe2)
#NULL
I could also report that plm(..., model="within", effect="twoways") is
quick when the number of individuals is kept low; re-run the same
example with n <- 200.
Any ideas on how to obtain the...