a suggestion on DLM instruction
a suggestion on DLM instruction
Dear Tom:
When we do a long loop for DLM instruction, we should skip non-convergent the result. We can use option=noprint, when it convergent, we can use option=print. However, DLM should support the save the output option and present the output for later without estimating. If possible, it can save the history of estimated results. Considering such situation, we do 100 times grid search, then according the convergent estimation, we only focus on and check the convergen estimation.
Best Regard
Hardmann
When we do a long loop for DLM instruction, we should skip non-convergent the result. We can use option=noprint, when it convergent, we can use option=print. However, DLM should support the save the output option and present the output for later without estimating. If possible, it can save the history of estimated results. Considering such situation, we do 100 times grid search, then according the convergent estimation, we only focus on and check the convergen estimation.
Best Regard
Hardmann
Re: a suggestion on DLM instruction
Can't you use tests on %CONVERGED to handle that? Just re-estimate with the PRINT option on when you get %CONVERGED==1.
Re: a suggestion on DLM instruction
Dear Tom:
we do grid search with two variables, and want to save the result to MS excel files. It seems only save the series with time spans defined by allocation.
Best Regard
Hardmann
we do grid search with two variables, and want to save the result to MS excel files. It seems only save the series with time spans defined by allocation.
Best Regard
Hardmann