Reasons for using the set.seed function

Question

Many times I have seen the set.seed function in R, before starting the program. I know it's basically used for the random number generation. Is there any specific need to set this?

Answer 1

The set.seed function is used to generate a reproducible random result.

For example:

Sampling without set.seed:

s1 <- sample(LETTERS,8)

> s1

[1] "Q" "E" "K" "R" "H" "O" "Y" "L"

> s1 <- sample(LETTERS,8)

> s1

[1] "X" "K" "Y" "J" "D" "P" "G" "I"
 

Sampling with set.seed:

> set.seed(123)

> s1 <- sample(LETTERS,8)

> s1

[1] "O" "S" "N" "C" "J" "R" "K" "E"

> set.seed(123)

> s1 <- sample(LETTERS,8)

> s1

[1] "O" "S" "N" "C" "J" "R" "K" "E"

Fixing the seed helps in optimizing a function that involves randomly generated numbers. If we do not fix the seed, the variation due to generating different random numbers will probably cause the optimization algorithm to fail.