RE: st: New variable based on a sum of products of another variable

[Nick Cox <n.j.cox@durham.ac.uk>]

Thanks for the update. At some point, it becomes natural to learn Mata for stuff like this. As they say, the learning curve is steep, meaning put in a little effort and you learn a lot. 

Nick 
n.j.cox@durham.ac.uk 

Seamer Paul

Many thanks to Nick and Paul for their suggestions,

Apologies that it’s taken me a while to reply.

Paul’s iterative solution is a neat answer to the question I posed, but there was an additional step that I didn’t mention in my original post,

I need to multiply each of the product terms (in parentheses in my original post) by another 2 variables before I perform the summation,

Essentially I’m looking generate a probability (P) that is of the form,

(P)= ∑ {var1} {var2} { ∏ var3 }

My original post related to calculating the sum product of var3, but I need to do this for different values of each variable and then perform a summation for each time period.

Nick's hint led me to a solution using logarithms ? it has a number of steps, but gives the answer I was looking for,
The values x are chronological years and i values are age in years (I'm using the probabilities to populate an illness-death Markov-type model).

.forvalues x = 1903/1997 {
.forvalues i = 1/41 {
.gen p`x'`i' = exp(sum(ln(var3`x')) - sum(ln(var3`x'[_n-`i']))) * var2`x'[_n-(`i'-1)]
.replace p`x'`i' = p`x'`i' * var1`x'[_n-`i'] if _n>=`i'+1
.}
.forvalues j = 2/41 {
.replace p`x'1 = p`x'1 + p`x'`j' if _n>=`j'
.}
.drop p`x'2-p`x'41 
.rename p`x'1 p`x' 
.}
 

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/