Hello,
I have a problem with categorical variables and dummy encoding. I've a
factor and
for each pair (i,j) with i != j, I'd like to fit
res ~ a*x[i] - b*x[j]. A brief example with 3 variables:
a - b = 2
b - c = -1
c - a = 0
Thus I fitted the following model:
fit <- lm(result ~ X + Y)
where Y is just the negative of X, means y[i] = -x[i].
But I don't want to double the variables. Thus I used the coding:
ht <- unclass(X)
at <- unclass(Y)
cnt<-0
for(j in 1:100) # number of equations
{
cnt<-cnt+1
y[cnt,at[j]]=1;
y[cnt,ht[j]]=-1;
}
fit <- lm(res ~ y[-length(y)]) # omit singularities.
>From the equations above
a - b = 2
b - c = -1
c - a = 0
a b c res
--------------
1 -1 0 2
0 1 -1 -1
-1 0 1 0
Since this matrix is singular. I omit last column (c)
and replaced it by intercept:
Change to
a b c(intercept) res
1 -1 1 2
0 1 1 -1
-1 0 1 0
and solved this. But somehow this incorrect.
I stuck in this since a couple of while. Does
anybody know how to solve this? Maybe solve.QP
is an answer to this?
Please help....
--
View this message in context:
http://n4.nabble.com/coding-negative-effects-in-categorical-dummy-variables-tp932611p932611.html
Sent from the R help mailing list archive at Nabble.com.
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
