Notice: On March 31, it was **announced** that Statalist is moving from an email list to a **forum**. The old list will shut down on April 23, and its replacement, **statalist.org** is already up and running.

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

From |
Eric Booth <ebooth@ppri.tamu.edu> |

To |
"<statalist@hsphsun2.harvard.edu>" <statalist@hsphsun2.harvard.edu> |

Subject |
Re: st: use random values in a calculation |

Date |
Mon, 4 Jul 2011 17:06:07 +0000 |

<> On Jul 4, 2011, at 9:52 AM, Zhi Su wrote: > The ratio to multiply one of the variables could be 0 too. The opt* variables created in my example already contained zero. Type -tab opt1- after it is created in my example to see this. (The -replace- commands in the opt* loop start with .05 because zero was already created as part of the -generate- command in that loop) > Probably > I do not ask the question right. The variables v1, v2, v3, v4 are > known. Yes, but not to me. So that why I included this part of my solution: >> >> **create fake v1-v4 variables for example: >> forv nn = 1/4 { >> g v`nn' = runiform()*10 >> } to create v1-v4. You can obviously omit this part if you've already got your data loaded into memory. > The ratios multiply them are chosen from (0,0.05,0.1,0.15,0.2). > Then each variable multiplied by a ratio are sum up. Totally, there > are 5*5*5*5=625 combination for these four variables multipled by a > ratio. Then I generate 625 variables for 625 types of sum. > In your original post you said that the ratios (opt*) are _randomly_ chosen to create sum 625 variables, which is what my example gave you...now it sounds more like you want every possible combination of the opt* and v1-v4 variables (?) [Though that's going to provide a lot more than 625 combinations] Here's an example for every v* x opt* combination: *********************** clear set obs 1000 // //changed from 625 obs g var = "s" + string(_n) //thanks to Nick Cox for pointing out this change **create fake v1-v4 variables for example: //skip this if you are using your real v1-v4 forv nn = 1/4 { g v`nn' = runiform()*10 } l in 1/9 //view fake v1 - v4 vars **multiply each v1 by each ratio: foreach var of varlist v1-v4 { foreach n1 in 0 05 1 15 2 { foreach n2 in 0 05 1 15 2 { foreach n3 in 0 05 1 15 2 { foreach n4 in 0 05 1 15 2 { g s`var'_`n1'_`n2'_`n3'_`n4' = (v1*`n1')+(v2*`n2')+(v3*`n3')+(v4*`n4') } //end n1 loop } //end n2 loop } //end n3 loop } //end n4 loop } //end var loop **also, you may want to check out -tuples- from SSC ************************ If this still doesnt answer your question, we need more information. You should (1) provide an example of what you want -- give us some example and/or hypothetical lines of data & variables that you have as well as some lines/variables that you want to see and (2) explain how/why either of the 2 examples do not solve your issue -- we can't guess what's not doing what you expect here (for instance, the first solution I gave provided 625 'sum' variables -- did you mean observations? did you expect to see the v1-v4 variables retained in the dataset? etc...) - Eric __ Eric A. Booth Public Policy Research Institute Texas A&M University ebooth@ppri.tamu.edu * * 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/

**References**:**Re:Re: st: use random values in a calculation***From:*Zhi Su <su.zh@husky.neu.edu>

- Prev by Date:
**Re: Re: st: binary mediation command** - Next by Date:
**st: Shapes of two Inverted U curves (same model on two different samples)** - Previous by thread:
**Re: Re: st: use random values in a calculation** - Next by thread:
**Re: st: use random values in a calculation** - Index(es):