Create a variable that identifies the original dat

问题:

I am relatively new to R and I would like to know how can I create a variable (number sequence) that identifies the each of the original data.frames before being joined with the rbind command.

Since in the original data frames there is one variable that is a row ID number, if creating a loop that assigns a new number in the new variable each time it encounters the number 1 in the row ID, it should work...

Thanks.

回答1:

There's a function in the gdata package called combine that does just that.

df1 <- data.frame(a = seq(1, 5, by = 1),
                  b = seq(21, 25, by = 1))

df2 <- data.frame(a = seq(6, 10, by = 1),
                  b = seq(26, 30, by = 1))

library(gdata)
combine(df1, df2)

    a  b source
1   1 21    df1
2   2 22    df1
3   3 23    df1
4   4 24    df1
5   5 25    df1
6   6 26    df2
7   7 27    df2
8   8 28    df2
9   9 29    df2
10 10 30    df2

回答2:

It looks like bind_rows from the dplyr package will do this too. Using maloneypatr's example:

df1 <- data.frame(a = seq(1, 5, by = 1),
                  b = seq(21, 25, by = 1))

df2 <- data.frame(a = seq(6, 10, by = 1),
                  b = seq(26, 30, by = 1))

dplyr::bind_rows(df1, df2, .id = "source")

Source: local data frame [10 x 3]

#    source     a     b
#     (chr) (dbl) (dbl)
# 1       1     1    21
# 2       1     2    22
# 3       1     3    23
# 4       1     4    24
# 5       1     5    25
# 6       2     6    26
# 7       2     7    27
# 8       2     8    28
# 9       2     9    29
# 10      2    10    30

回答3:

Why not just:

    rbind( cbind(df1, origin="df1"),
           cbind(df2,  origin='df2') )

Or if you want to preserve rownames:

  rbind( cbind(df1, origin=paste("df1",rownames(df1), sep="_") ),
         cbind(df2, origin=paste("df1",rownames(df1), sep="_") ) )

回答4:

You can use

transform(dat, newCol = cumsum(ID == 1))

where dat is the name of your data frame and ID is the name of the ID column.

回答5:

A fairly extensible solution:

# test data:
df1 <- data.frame(id=letters[1:2])
df2 <- data.frame(id=letters[1:2])

Collect your data into a list then rbind all at once:

dfs <- c("df1","df2")
do.call(rbind, Map("[<-", mget(dfs), TRUE, "source", dfs) )

#      id source
#df1.1  a    df1
#df1.2  b    df1
#df2.1  a    df2
#df2.2  b    df2

Also, note in this example that when you rbind using a named list, your rownames reference the source data. This means you can nearly get what you want using just:

dfs <- c("df1","df2")
do.call(rbind, mget(dfs) )

#      id
#df1.1  a
#df1.2  b
#df2.1  a
#df2.2  b

回答6:

Thanks all! I ended up with a simple solution working with a friend of mine by creating an index, like this:

index<-rep(1,times=nrow(data.frame))

for (i in 1:(nrow(data.frame)-1)){

if (data_frame$ID [i+1]<= data.frame$ID[i]) {
index[i+1]<-index[i]+1
}
else {index[i+1]<-index[i]}}

new.data.frame <- cbind(index, data.frame)