I have df below.
df<-cbind(c("LA", "NY", "Rome"),c(1,1,2),c(0,1,0),c(0,1,2),c(1,3,4))
[city] [c1] [c2] [c3] [sum]
[1,] "LA" "1" "0" "0" "1"
[2,] "NY" "1" "1" "1" "3"
[3,] "Rome" "2" "0" "2" "4"
I want to compare the values in sum column to find the two smallest values .( here 1,3 in increasing order)
then return the corresponding values in city column with the same order.
so what I want in out put is : LA, NY
I can get the rownames ( which I do not want, but I do not know how to get LA, NY:
rownames(df$city[order(df$sum, decreasing = F),][1:2])[1:2]
The dataset in the question is a matrix (cbind returns a matrix by default and matrix can have only a single type. Thus, the whole dataset is converted to character). If we need to extract 'n' elements of first column based on the last column (5th column), extract the 5th column, convert to numeric, use order to get the index of values in ascending order, then with head get the n order index and use that to subset the first column
n <- 2
df[,1][head(order(as.numeric(df[,5])), n)]
[1] "LA" "NY"
as.data.frame(cbind is wrong way as the columns are already getting type changed to character. It is better to use data.frame directly
df1 <- data.frame(col1 = ..., col2 = .., col3 = ...)