I'd call this standardisation, not normalisation. Your problem is
essentially that of Paul Byatta posted yesterday
http://www.stata.com/statalist/archive/2013-04/msg01088.html
It seems that you have three variables -x1-, -x2-,-x3- and various -Period-s.
forval i = 1/3 {
egen mean = mean(x`i'), by(Period)
egen sd = sd(x`i'), by(Period)
gen N_x`i' = (x`i' - mean) / sd
drop mean sd
}
should help.
The problem with your code is the lines
qui sum `x' if Period==`i', meanonly
replace sd_`x'=r(sd) if Period==`i'
With the -meanonly- option -summarize- does not calculate the sd.
When you refer to r(sd) afterwards, Stata just substitutes missing,
and this is propagated to all your new variables.
But as above, you don't need so much code.
Nick
njcoxstata@gmail.com
On 25 April 2013 18:53, Xixi Lin <winnielxx@gmail.com> wrote:
> I am trying to normalize variables, and the following is my code:
> foreach x in x1 x2 x3{
> gen N_`x' = .
> gen mean_`x'=.
> gen sd_`x'=.
> }
> forvalues i=2/120{
> foreach x in x1 x2 x3{
> qui sum `x' if Period==`i', meanonly
> replace mean_`x'=r(mean) if Period==`i'
> qui sum `x' if Period==`i', meanonly
> replace sd_`x'=r(sd) if Period==`i'
> }
> }
>
>
> forvalues i=2/120{
> foreach x in x1 x2 x3{
> replace N_`x'=(`x'-mean_`x')/sd_`x' if Period==`i'
> }
> }
>
> however, the above code does not working, because it creates all
> missing values for standard deviation. Can anyone help me to fix the
> code?
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/
