Why?
So far we have looked at statements that were executed when a condition was fulfilled, but nothing happened when it was not:
# input has to be number
input <- 12
if(input>=0) result <- "Awesome!"
message(result)
Awesome!
In this case, if we enter a negative value as input, then nothing at all will happen, which can have dire consequences:
# input has to be number
input <- -52
if(input>=0) resultNew <- "Awesome!"
message(resultNew)
Error: object 'resultNew' not found
This error occured because our workspace does not have a resultNew object - it would only be defined if the value of input>=0 were TRUE, but it was FALSE.
To make this program work with input set to a negative value, we would have to provide a complimentary expression, that makes sure that the variable we define exists:
# input has to be number
input <- -52
if(input>=0) resultNew2 <- "Awesome!"
if(input<0) resultNew2 <- "Not great!"
message(resultNew2)
Not great!
Such complementary needs are very frequent in programming languages. To make them simpler, they are usually expressed with if-else paired statements.
# input has to be number
input <- -52
if(input>=0) resultNew2 <- "Awesome!" else resultNew2 <- "Not bad!"
message(resultNew2)
Not bad!
In our case this chunk of code behaves exactly the same as the one above - but is also faster, and less error prone, as you only have to evaluate one logical expression: input>=0. Also note, that instead of two potential forking points, you have just one (with the double if case you actually have the ability to alter the value of input, and then the logical statements might not be complementary anymore!).
The blueprint for this is the same as with simple if statments, but with the addition of the else keyword:
if(<logical value>) <commandWhenTRUE> else <commandWhenFALSE>
You can substitute the single commands with entire blocks (sets of statements between braces: {}). The following code is exactly the same as the one before.
# input has to be number
input <- -52
if(input>=0){
resultNew3 <- "Awesome!"
}else{
resultNew3 <- "Ooh, even better!"
}
message(resultNew2)
Ooh, even better!
The use of if-else pairs is extremely common in R, usually more common then simple if statements.