error in user defined function which generate statistics by group

Question

I'm trying to make function to return statistics by group like followings. To make statistics by group, I used subset in this function code. But errors occurred when argument 'y' is applied to 'subset'. How can I solve this problem? I'll wait wisdom of yours. It's sure that tapply can be used, but my intention is to make function. Thank you.

sbyg<-function(dt,grp,y) {
# dt=data.frame, grp=group variable, y=value variable
ng<-length(unique(grp))
x<-as.vector(unique(grp))
statis<-matrix(nrow=ng,ncol=6)
for (i in 1:ng) {
  dta<-dt[grp==x[i],]
  attach(dta)
  statis[i,1]<-nrow(dta) # 건수
  statis[i,2]<-colSums(!is.na(dta))[1] # nonmiss건수
  statis[i,3]<-mean(dta[,y],na.rm=TRUE) # 평균
  statis[i,4]<-median(dta[,y],na.rm=TRUE) # 중위수
  statis[i,5]<-min(dta[,y],na.rm=TRUE)
  statis[i,6]<-max(dta[,y],na.rm=TRUE)
  detach(dta)
}
rownames(statis)<-x
colnames(statis)<-c("count","nonmiss","mean","median","min","max")
statis
}

sbyg(iris,Species,Sepal.Length)  # error occurs

You can use one of the group by functions to do this For example with `aggregate` or `data.table` or `ddplyr` — akrun, Feb 04 '16 at 11:20
idem as [here](https://stackoverflow.com/questions/9847054/how-to-get-summary-statistics-by-group) — YCR, Feb 04 '16 at 11:24

YCR · Answer 1 · 2016-02-05T09:29:02.690

Your call of the function should be this:

sbyg(iris,"Species","Sepal.Length")

unless iris is a data.table object (which is not by default).

edit: Modified function:

sbyg<-function(dt,grp,y) {
  # dt=iris ; grp="Species"; y="Sepal.Length"
  ng<-length(unique(dt[, grp]))
  x<-as.vector(unique(dt[, grp]))
  statis<-matrix(nrow=ng,ncol=6)
  for (i in 1:ng) { # i <- 1
    dta<-dt[dt[, grp]==x[i],]
    statis[i,1]<-nrow(dta) # 건수
    statis[i,2]<-colSums(!is.na(dta))[1] # nonmiss건수
    statis[i,3]<-mean(dta[,y],na.rm=TRUE) # 평균
    statis[i,4]<-median(dta[,y],na.rm=TRUE) # 중위수
    statis[i,5]<-min(dta[,y],na.rm=TRUE)
    statis[i,6]<-max(dta[,y],na.rm=TRUE)
  }
  rownames(statis)<-x
  colnames(statis)<-c("count","nonmiss","mean","median","min","max")
  statis
}

But this is not an optimal function. The best way is to use tapply.

Thank you for your concise answer, YCR quotation mark(") almost solved my problem. but, 'by group' doesn't work. I wonder what's the problem. — hyunwoo jeong, Feb 05 '16 at 00:34

error in user defined function which generate statistics by group

1 Answers1