CODEWITHSUNDEEP
r
user5316628
user5316628

Reputation: 357

R return column name based on conditions

For the following data frame:

DF <- data.frame(Row=c(1,2,3,4,5),`2.04`=c(1,1,0,1,1),`2.05`=c(0,0,0,0,1),
       `2.06`=c(1,0,0,0,1),`2.07`=c(1,0,0,0,1),`2.08`=c(1,1,1,0,0), check.names = F)

I'd like to return into a new vector for each row the column name that has both both (a) a value greater than 0 in the relevant row; (b) column name has the highest value of all those that meet condition a, such that:

DF <- data.frame(Row=c(1,2,3,4,5),'2.04'=c(1,1,0,0,1),'2.05'=c(0,0,0,0,1),
                '2.06'=c(1,0,0,0,1),'2.07'=c(1,0,0,1,1),'2.08'=c(1,1,1,0,0),
                Results=c(2.08,2.08,2.08,2.04,2.07)

So for row 2 the columns 2.04 and 2.08 meet condition (a), and only 2.08 meets condition (b) because 2.08>2.04.

dplyr or data.table would be preferred.

Upvotes: 2

Views: 10339

Answers (2)

lmo
lmo

Reputation: 38520

You could also use max.col like this

DF$results <- names(DF[-1])[max.col(DF[-1], "last")]
DF
  Row 2.04 2.05 2.06 2.07 2.08 results
1   1    1    0    1    1    1    2.08
2   2    1    0    0    0    1    2.08
3   3    0    0    0    0    1    2.08
4   4    1    0    0    0    0    2.04
5   5    1    1    1    1    0    2.07

max.col returns the column position of the maximum value for each row. It takes a second argument, ties.method, which is set to "last" here in order to return the largest column position for each row. These column positions are used to extract the column names with [ which are then converted to numeric and put into a vector.

Upvotes: 4

Ronak Shah
Ronak Shah

Reputation: 389325

We can use apply row-wise and get the names for all the columns whose value is more than 0 and get the max of it. 

DF$Results <- apply(DF[-1], 1, function(x) max(names(which(x >0))))

DF
#  Row 2.04 2.05 2.06 2.07 2.08 Results
#1   1    1    0    1    1    1    2.08
#2   2    1    0    0    0    1    2.08
#3   3    0    0    0    0    1    2.08
#4   4    1    0    0    0    0    2.04
#5   5    1    1    1    1    0    2.07

Upvotes: 3

Related Questions