Hi Gianluca,

Thank you for your prompt comment.

>>sig_eps = 0.5

>>y = rnorm(n, mu, sig_eps)

Do you know how to generate the value of sig_eps from the data without initial setting value?

Could you please advice any formula to generate sig_eps ? 

I may be mistaken, but I think Aisyah has raw data, and wants to know how to compute (and maybe plot?) the residuals from a linear regression model (using R).  If so, try a Google search on something like
.  You should find plenty of examples, like the one linked below.  HTH.

http://www.r-tutor.com/elementary-statistics/simple-linear-regression/residual-plot

I think the R code will still be useful, even if the real question is about real data. If you do not understand the outcome that is generated based on rules that you set up, then maybe you should get help. However, the R code could use some annotation. The short version is that you are setting up a linear dependence of y on x such that y=1+0.25*x. You are then imposing a random element onto y using a random number generator based on the normal distribution. The result, with a large sample size, will be indistinguishable from normally distributed. So, with the program provided you need to answer a few questions.

1) How does the distribution change as n changes? What if n=10, or n=10000?

2) How does the distribution change as sig_eps change?

3) Can you find random number generators in R that will produce different distributions? A Google search for R and random number might be instructive.

Do you have y and x? Then

e

Hi Aisyah,

Could you please show a sample from your data?

You may use:

head(dataa) to show the first six rows in your data. Also, please identify the Dependent and independent variables.

For linear regression in R, you can use lm function:

lm

In my opinion the simulation (generation) of the random error that is normally distributed should be implemented taking into account the prediction interval of regression, which, for example, according to Ryan. (1997) depends not only on the prediction linear regression equation result and the standard deviation of the of sample of dependent variable but also from the quantile of student's t-distribution and the number of values of dependent variable.

Ryan, T.P. (1997). Modern regression methods. Wiley, 529 p. ISBN 0471529125