Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: -rqdeco- and -cdeco- help

From   Jorge Eduardo Pérez Pérez <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: st: -rqdeco- and -cdeco- help
Date   Thu, 18 Aug 2011 13:23:17 -0400

You can do several things:

* Reduce the number of quantiles estimated. The default is 100, you
can use 50, or 10.
* Reduce the number of bootstrap replications
* Type -set matafavor speed-
* Run two sets of the bootstrap in parallel sessions. This is a little
involved, and I have not tried with bootstrapping myself, but I think
you can adapt the guidelines here:

to run your bootstrap replications in several sets, and then merge
them back. I have tried the example with multiple imputation  posted
in that website and I obtain speed improvements against running the
whole imputation in just one session of Stata. Have not tried with
bootstrapping, but it should work the same way.

I can provide some more guidelines if you want to try the last method,
but my advice would be to try the other suggestions first.

Jorge Eduardo Pérez Pérez

On Thu, Aug 18, 2011 at 4:43 AM, Muhammad Anees <[email protected]> wrote:
> Hi
> I have been trying to decompose earnigs function between two groups
> using the two stata user written code, namely -rqdeco- by Blaize Melly
> and -cdeco- by Ivan Fernandez. I am unable to have any results even
> waiting for 10-15 hours leaving my laptop to work the whol nights, but
> no success. I am using Stata 10 on Windows 7, 32B and 2GB ram and 2GH
> processor. Is there any idea which can help me to obtain the results
> in shorter times. Or any other alternatives, I would be thankful.
> The codes could be downloaded from
> -cdeco- from
> -rqdecom from
> --
> Thanks
> Anees
> *
> *   For searches and help try:
> *
> *
> *

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index