Let X1,...,Xn be a random sample from the pdf

f(x)=(1/b)exp[-(x-a)/b], a< x< infinity, 0< b< infinity,

where a and b are location and scale parameter, respectively.

Now I have to generate X of 1000 observations from the pdf. I am a R user.

For initiate the problem, I set parameter values of a and b arbitrarily. But to generate the probability f(x), I need to set the values of x, but I have been asked to generate X from the pdf. It seems to calculate the probabilities f(x) and to generate the values of X are simultaneous work. How can I do that ?

  • $\begingroup$ (Trying not to give a complete solution to a homework problem:) The usual method to take a first stab at generating random variables (if prevented from using a built-in function) from a particular distribution is to invert the function to get a quantile function and then feed it random variates from the uniform distribution on the interval [0,1]. You can look at the shape of the distribution thereby produced using hist. hist( -log( runif(1000) ) ). I'm guessing this is homework from a mathematical statistics course and that you should demonstrate steps in getting to your final solution. $\endgroup$ – DWin Nov 30 '15 at 2:19
  • 1
    $\begingroup$ Close vote on the hypothesis that this is request of stats advice. $\endgroup$ – DWin Nov 30 '15 at 2:24
  • $\begingroup$ @42- hm, that's inversion method. $\endgroup$ – user 31466 Nov 30 '15 at 2:44
  • $\begingroup$ Yes, it is. What is your point? $\endgroup$ – DWin Nov 30 '15 at 2:45

Well, that looks like exponential distribution

Use rexp function to generate vector with rate equal to 1/b and then shift the result by adding a.

Link: https://stat.ethz.ch/R-manual/R-devel/library/stats/html/Exponential.html

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.