DATA Step, Macro, Functions and more

monte carlo simulation with summation

Occasional Contributor
Posts: 19

monte carlo simulation with summation

I'm trying to do a monte carlo simulation where I need to sum over the total data set as a last step and can't seem to make it work, let alone do it in a nice way. 


I want to simulate the variable A that is made from a number of parameters and the standard normal distribution. From this I want to sum another variable w if A fulfills a specific requirement.


My problem is firstly that i can't find a way to sum these w in the same data step as the simulation of A. Since my data set is about 10 000 rows and i want to do 500 000 simulations I can't have the table growing to 500 000 * 10 000 but rather i want the final table to have the sums of each iteration, in other words it should be 500 000  rows 


Furthermore i'm interested in if there is a more elegant way of approaching this problem in terms of the functions used etc.


I'm using SAS 9.3 if its relevant.




this is what i have so far, but it is not working...


data multi;
set multi_base;

do iter = 1 to 2;

L = 0;

A = a1*rand("Normal") +
a2*rand("Normal") +
a3*rand("Normal") +
a4*rand("Normal") +

if A < b then L = L + w;

keep iter L;

Posts: 1,400

Re: monte carlo simulation with summation

Posted in reply to jan_t_lagen

2 questions: 


Do you have a SAS/IML license?


What does the data set multi_base look like?

Occasional Contributor
Posts: 19

Re: monte carlo simulation with summation

I do not have a SAS/IML licence and multi_base just contains a1-a5, b and w
that are unique for each row
Posts: 4,267

Re: monte carlo simulation with summation

Posted in reply to jan_t_lagen

I assume that if A >= b then L remains unchanged and that value of w is discarded?



Compute the conditional cumulative sum as you go along, but put the OUTPUT statement OUTSIDE the END of the iteration loop. 



Ask a Question
Discussion stats
  • 3 replies
  • 3 in conversation