Statalist


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

st: RE: RE: RE: Data generating


From   "Torben Mideksa" <torben.mideksa@cicero.uio.no>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: RE: RE: Data generating
Date   Sat, 22 Nov 2008 20:00:31 +0100

Thank you Nick, I have now fixed it. I like the trick "If the requirement is
just non-zero variance then one solution is just to add random noise with
desired distribution and specified variance (Presumably also with zero
mean.)"

Best Regards,
Torben

>-----Original Message-----
>From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-
>statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
>Sent: 21. november 2008 13:22
>To: statalist@hsphsun2.harvard.edu
>Subject: st: RE: RE: Data generating
>
>I guess many people found the question a bit obscure -- or the answer
>that occurred to them seemingly too obvious that it seemed that someone
>else would say it.
>
>If the requirement is just non-zero variance then one solution is just
>to add random noise with desired distribution and specified variance.
>(Presumably also with zero mean.)
>
>But it is difficult to see how that could be of scientific or
>statistical interest. What is missing here is any information on
>seasonality, i.e. whatever your data, economic, biomedical, something
>else, there is likely to be an expectation of seasonality, and only you
>can say how that is best to be imputed when cooking up a monthly series
>from annual data.
>
>Remember we know nothing about your data and your problem that you
>haven't told us. In fact, I don't think there is anything substantive
>here to use as leverage in advice. There can be hardly be a trick
>regardless of context to impute seasonality, as seasonality takes many
>different forms.
>
>Nick
>n.j.cox@durham.ac.uk
>
>Torben Mideksa
>
>10Q Carlo!
>However, what I needed is to generate monthly data for 50 or more years
>for more than 40 cases. Unless I am missing something from your answer,
>while the approach you suggested works focusing on a single year for a
>single case, it is awkward to work with for a large data set with many
>years and many cases. I am wondering if there is a trick in stata that
>generates the monthly data at once I feed the annual data for all cases.
>I can live with zero variance property of the monthly data for a while
>if non-zero variance results in huge computational cost. Once again,
>thanks ahead of time for your help.
>
>Carlo Lazzaro
>
>Question nr 1: How do I do this in stata?
>>Answer nr 1 (please, consider that what follows refers to each one of
>>your years. I assume that the annual value a is 100):
>>set obs 12
>>g a=100/12
>>Question nr 2: How do I do it in stata if I want to assume a non-zero
>>variance still maintaining the sum of the monthly data equals the
>>annual data?
>>Answer nr 2: corr2data b, means(8.3333333)
>>
>>sum  a b
>>
>>    Variable |       Obs        Mean    Std. Dev.       Min        Max
>>-------------+--------------------------------------------------------
>>           a |        12    8.333333           0   8.333333   8.333333
>>           b |        12    8.333333           1   6.919311   9.956305
>>
>>. total a b
>>
>>Total estimation                    Number of obs    =      12
>>
>>--------------------------------------------------------------
>>             |      Total   Std. Err.     [95% Conf. Interval]
>>-------------+------------------------------------------------
>>           a |        100          0             .           .
>>           b |        100   3.464102      92.37556    107.6244
>>--------------------------------------------------------------
>
>Torben Mideksa
>
>>I have a list of annual data like
>>
>>Year		variable_X		variable_Y ...
>>1900		a			b
>>.		.			.
>>.		.			.
>>.		.			.
>>1950		c			d
>>
>>For some reasons, based on annual data alone, I am interested in
>>generating monthly data for the two or more variables, assuming the
>>monthly data has zero variance with the average value for each month
>>amounting to annual value divided by 12. That is, I need
>>
>>Year		variable_X		variable_Y ...
>>1900M1		a/12			b/12
>>1900M2		a/12			b/12
>>1900M3		a/12			b/12
>>.		.			.
>>.		.			.
>>.		.			.
>>1900M12		a/12			b/12
>>.		.			.
>>.		.			.
>>.		.			.
>>1950M12		c/12			d/12
>>
>>Question nr 1: How do I do this in stata?
>>Question nr 2: How do I do it in stata if I want to assume a non-zero
>>variance still maintaining the sum of the monthly data equals the
>>annual data?
>
>
>*
>*   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/

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index