I have a presumably simple question. What does this mean: with probability 0.3 variable X is simulated as
X~N(0,1).
How can I simulate X using this?
$\begingroup$
$\endgroup$
5
-
1$\begingroup$ What happens if the N(0,1) does not happen? A deterministic zero? $\endgroup$– Stephan KolassaFeb 20, 2013 at 14:26
-
$\begingroup$ Simulate... what? The 1s and 0s that might make up X? $\endgroup$– russellpierceFeb 20, 2013 at 14:31
-
$\begingroup$ Well thank you both for your interest! Below richard gave exactly the point i couldnt grt. Thank you. $\endgroup$– ChristosFeb 20, 2013 at 15:12
-
$\begingroup$ It sounds like the analyst only has a 30% chance of being at work today. :-) More seriously, the phrase is nonsensical because it is silent about what happens the other 70% of the time--and by its construction it (bizarrely) suggests the possibility that 70% of the time a the variable won't be simulated at all! $\endgroup$– whuber ♦Feb 21, 2013 at 8:55
-
$\begingroup$ Well the exact phrase is: simulated data were generated as follows. With prob '0.3 X~N(0,1)' and with prob '0.7 X~N(y,1)' (while y is known). $\endgroup$– ChristosFeb 22, 2013 at 11:52
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
What about first sampling a uniform $U$. If $U<0.3$ (this has probability $0.3$) then you sample $N(0,1)$ else you do something else.
