randomization process coding

[Ali Niazi]

Hi Everyone,

I have 13 events, and I want to have a probability for each.

- Define the Partial_Sum(j)=sum(p(k) to p(j)), where j=1,2,...13 with Partial_Sum(14)=1
- To sample from this discrete distribution, a random variable X is drawn from a continuous uniform distribution on the interval (0,1)
- For each value of j check which of the following N mutual events is true:
* X<=Partial_Sum(1)
* Partial_Sum(j-1)<X<=Partial_Sum(j) for j=2,3,...,13

I need to write a code for this. Could you please help me with this?

[Clyde Schechter]

I've illustrated an approach with my own made up set of 13 partial sums.

* Example generated by -dataex-. For more info, type help dataex
clear*
input int index float partial_sum
 1 .14785433
 2  .1692771
 3 .19177543
 4 .19815916
 5 .21698804
 6  .2227877
 7  .2450673
 8 .25898004
 9  .6292577
10  .7572148
11  .8604344
12  .8686824
13         1
end
tempfile partial_sums
save `partial_sums'

//    SIMULATE 100 DRAWS FROM THE PDF DEFINED IN `partial_sums'
clear
set seed 1234
set obs 100
gen x = runiform()
gen int obs_no = _n

cross using `partial_sums'
by obs_no (index), sort: egen wanted = min(cond(partial_sum >= x, index, .))
keep obs_no x wanted
duplicates drop

[Ali Niazi]

Thank you so much, Clyde!