I have a non-linear optimization problem to solve. Specifically, I want to find scale and shape paramters that maximize the log likelihood for Weibull distribution for interval-censored time to failure/replacement observations. I have a piece of code using PROC NLP that solves this problem. The dataset containing the time to failure data has lifetime data for various components. Weibull distribution is to be fit at the component level. Hence Proc NLP uses a By statement. We have something like this:
PROC NLP DATA=<>;
What is the equivalent construct for BY statement in PROC OPTMODEL?