Using ifelse in R

I am trying to code the following statement in R with if and ifelse.The sample data is trial and x,y,and z are columns of trial).

Statements to be coded

if (x>0) {
      if (y>0) {
          l=2
       }else{
           l=5
       }
      if (z>0) {
          m=l+2
       }else{
           m=5
       }
}

The R code using ifelse

trial$l<-with(trial, ifelse((x>0 &y>0),2,ifelse((x>0 &y<=0),5,???)))
trial$m<-with (trial,ifelse((x>0 &z>0),l+2,ifelse((x>0 &z<=0),5,???)))

where, ??? specifies that there are no values according to the above statement. In other words for x<0 and y there are no values.

Next, I use combination of if and ifelse to see that works:

if(trial$z>0){
    trial$l<-with(trial, ifelse(y>0,2,5))
    trial$m<-with(trial, ifelse(z>0,l+2,5))
}

This code is ok but there is a warning message (since z is a column vector)

In if (trial$z>0){
the condition has length>1 and only the first element will be used

I want to focus only on using ifelse since I am dealing with only vector. But, I have no luck in this regard. Any idea?

Solution 1:

If you want to use ifelse and nest things you could do something like this

test <- data.frame(x = 2, y = 5, z = 3)
with(test, ifelse(z > 0 & x > 0 | y > 3, "yes", "no"))

In this case you're using logical operators to guard the output. You'll still get "no" if z <= 0, but you can deal with that pretty easily.

with(test, ifelse(z > 0, ifelse(x > 0 | y > 3, "yes", "no"), NA))

Nested ifelse statements can get hard to follow in any language, so consider matching or switch statements if you end up with more than 3 of them.

Solution 2:

I would use transform twice for example:

trial <- data.frame(x=c(-1,1,2),y=c(1,-2,3),z=c(1,-5,5))
trial <- transform(trial,l = ifelse(x>0,ifelse(y > 0,2,5),NA))
transform(trial,m = ifelse(x>0,ifelse(z>0,l+2,5),NA))

   x  y  z  l  m
1 -1  1  1 NA NA
2  1 -2 -5  5  5
3  2  3  5  2  4

Note that I assign NA for case x < 0. You can use a one transform like this for example:

trial <- data.frame(x=c(-1,1,2),y=c(1,-2,3),z=c(1,-5,5))
transform(trial,l <- ifelse(x>0,ifelse(y > 0,2,5),NA),
                         m = ifelse(x>0,ifelse(z>0,l+2,5),NA))
  x  y  z c.NA..5..2.  m
1 -1  1  1          NA NA
2  1 -2 -5           5  5
3  2  3  5           2  4

But personally I would prefer the first one for readability besides the fact you need maybe change column names.