Exequiel and Nazmul already gave you the theory. Now if you want a straightforward solution, follow my example.

You have something like this:

--------------------------------------------------------------------

madeup_data

Do you have the standard deviation of errors? If you don't, you need to assume one value.

From the general equation of linear regression, with an error term e, which follows a normal distribution N(0,sigma_e):

y = mx+n + e

you can simulate y values as follows, assuming that x is in the range of (0,999):

x = runif(100, min = 0, max = 999)

y = m*x+y + rnorm(100, mean = 0, sd = sigma_e)

Saurav Vaia,

The mean is the regression equation itself. However, the std. dev. can/may be back calculated from the R-square value.

Exequiel and Nazmul already gave you the theory. Now if you want a straightforward solution, follow my example.

You have something like this:

--------------------------------------------------------------------

madeup_data