CODEWITHSUNDEEP
r
Joram
Joram

Reputation: 159

How to add a row to a dataframe and avoid "non-numeric argument to binary operator"

I have a dataframe. I want to normalize columns 2 and 3 by dividing them by the maximum value of column 2 and 3. 

> testdf<- data.frame("a"=c("b",2), "b"=2:3, "c"=3:4, "d"=4:5, stringsAsFactors = F)
> testdf
  a b c d
1 b 2 3 4
2 2 3 4 5

> testdf[2:3]<-testdf[2:3] / do.call(pmax, testdf[2:3])
> testdf
  a         b c d
1 b 0.6666667 1 4
2 2 0.7500000 1 5

Notice how the df contains a mix of numerical and string values? Now I want to add a row with more data. If the first element of the added row is a string, the code gives an error. 

> testdf<- data.frame("a"=c("b",2), "b"=2:3, "c"=3:4, "d"=4:5, stringsAsFactors = F)
> testdf
  a b c d
1 b 2 3 4
2 2 3 4 5
> testdf<- testdf %>% rbind(c("a",6,7,8))
> testdf
  a b c d
1 b 2 3 4
2 2 3 4 5
3 a 6 7 8
> testdf[2:3]<-testdf[2:3] / do.call(pmax, testdf[2:3])
Error in FUN(left, right) : non-numeric argument to binary operator

If instead I use only numerical values, it works. 

> testdf<- data.frame("a"=c("b",2), "b"=2:3, "c"=3:4, "d"=4:5, stringsAsFactors = F)
> testdf
  a b c d
1 b 2 3 4
2 2 3 4 5
> testdf<- testdf %>% rbind(c(5,6,7,8))
> testdf
  a b c d
1 b 2 3 4
2 2 3 4 5
3 5 6 7 8
> testdf[2:3]<-testdf[2:3] / do.call(pmax, testdf[2:3])
> testdf
  a         b c d
1 b 0.6666667 1 4
2 2 0.7500000 1 5
3 5 0.8571429 1 8

Any help to why this happens is greatly appreciated. I need to be able to add rows that contain text and numbers while keeping the code working. My guess is that I'm messing up types but I couldn't figure out, how.

Upvotes: 1

Views: 209

Answers (2)

d.b
d.b

Reputation: 32548

When you do rbind(c("a",6,7,8)) you are effectively doing rbind(c("a","6","7","8")) thereby making everything in testdf character. This is because a vector (c(...) or individual columns of testdf) can hold data of only one type and R will try to do so while accommodating all data. In this case, character would store all data but numeric would get rid of the letters for example.

Just use testdf %>% rbind(list("a",6,7,8)) instead of testdf %>% rbind(c("a",6,7,8)).

Compare the output of list("a",6,7,8) vs that of c("a",6,7,8).

Upvotes: 1

akrun
akrun

Reputation: 887008

We can use add_row

library(tibble)
testdf <- add_row(testdf, !!!set_names(list('a', 6, 7, 8), names(testdf)))
testdf
#  a b c d
#1 b 2 3 4
#2 2 3 4 5
#3 a 6 7 8

Now, do the pmax on the numeric columns 

testdf[2:3] / do.call(pmax, testdf[2:3])
#          b c
#1 0.6666667 1
#2 0.7500000 1
#3 0.8571429 1

Upvotes: 1

Related Questions