rbind error: "Error in match.Names(clabs, Names(xi))"

Question

I’m trying to build a function that counts the number of complete cases in files. csv (data frames), ie the number of lines with values (and not "NA"). The function I wrote can read the file(s) specified in the function call and count the number of rows with complete cases, but I need to return this information as a data frame, with 2 columns, "id" for the file identification (they are all identified with numbers) and "Nobs" for the number of observations (the complete cases). In this function, my read loop cannot then store the count results in that data frame. The figure below has some examples of what the function should return.

That’s the job of:

completeF<-function(directory,id) {

#set file location
address<-paste(getwd(), directory, sep="/")

#Creates objects for later results keeping
Completev<-data.frame(matrix(0,1,ncol=2))
temp<-data.frame(matrix(0,1,ncol=2))
#colnames(Completev)<-c("id","Nobs")
#colnames(temp)<-c("id","Nobs")

#read files
files <- dir(directory)

for (i in id){
    #read files
    each.file<-read.csv(paste(address,files[i],sep="/"),h=TRUE)

    #count complete cases: count number of lines with values for sulfate and nitrate
    obs<-na.omit(each.file)
    rowcount <- nrow(obs)

    #keep results in temporary data frame, and then the final one
    temp<-cbind(i,rowcount)
    Completev<-rbind(Completev,temp)
}
colnames(Completev)<-c("id","Nobs")
return(Completev)

}

But what I get is

> completeF("specdata",1:5)
Error in match.names(clabs, names(xi)) : 
  names do not match previous names 

Running Traceback, I understood that the error occurs here: Completev<-rbind(Completev,temp), with the rbind, but I don’t understand why! Even fixing the names of the columns before the "for" loop, it solves.

What could be wrong in the loop?

Answer 1

You are reassigning the variable temp with each iteration and deleting the column names.

One option is to copy the column name at each iteration:

temp<-data.frame(i,rowcount) # nova data frame com seus próprios nomes
names(temp) <- names(Completev) 
Completev<-rbind(Completev,temp)

You can also create a new data frame already with the names to each interaction:

Completev<-rbind(Completev,data.frame(id=i,Nobs=rowcont)

Note however that in terms of performance this is not a good solution, it is much more efficient to preallocate a vector and build the data.frame at once (see that question on Soen for some benchmarks).