提取R中的自由度

Question

I am running a large number of linear regressions, and for each regression I would like to save the adjusted R squared and the degrees of freedom each in a seperate file.

The code below does this perfectly for the adjusted R squared, and I can add the value name of the list to the file (so I can identify to which subset of mydata the R squared belongs). In sum, it gives me 1 column with the list value, and 1 column with the R².

For the degrees of freedom, R gives a 3-vector (p, np, p*) by default. Adding the row.name=i statement does not work here. I would only need np. Is there a way to save the middle figure from the vector only and add the row name to the saved file?

I created a reproducible example below:

Fictive dataset (mydata)

v1  v2  v3

4   9   1
5   78  1
6   11  1
8   12  2
10  13  2
11  65  2
15  34  2

Code:

list<-unique(mydata$v3)
for (i in 1:length(list)){


  newdata <- mydata[ which(mydata$v3==i),]

  reg <- lm((v1~v2), data=newdata)
  write.table(summary(reg)$adj.r.square, file="rsquared.txt", append=TRUE, col.names = F, row.names=i)
  write.table(summary(reg)$df, file="degreesoffreedom.txt", append=TRUE, col.names = F)


  rm(reg)
  rm(newdata)

}

What I currently get for the file "degreesoffreedom.tx"

"1" 2
"2" 1
"3" 2
"1" 2
"2" 2
"3" 2

What I need is:

"1" 1
"2" 2

Answer 1

Try summary(reg)$df[2] instead of summary(reg)$df . Remember everything in R is an object and you can take a subset of any object.

Second solution reg$df . You'll only get three different DF when you wrap summary around your regression.

Code suggestion

The call to which is unnecessary, ie newdata <- mydata[mydata$v3==i,] . In my experience, 99% of the time you think you need which you don't.