R help - Sep 2005 - Searching for specific functions

Hello everybody (I'm new here)

  I am looking for 2 specific functions. The first one is a variance
formula that does not divide by n-1 (I don't want the unbiased formula,
that's it). In case it does not exist, how do I modify the default
function code? Yes, a workaround would be to multiply the result by n-1
and then divide it by n, but I'd like to use a separate function, if
possible.

  The other function I'm looking for it's a median formula like this
one:

http://www.faberitius.com/prototipo/gfx/clip_image002_0000.gif

  Where "Li" is the exact lower limit of the critical interval,
"nb" is
the number of cases that lie below the critical interval, "nd" the
number
of cases inside the mentioned interval, and "I" is the range of the
interval itself. As you can see this formula calculates the median in
categorical structured data.

  I couldn't find any functions that match the results done by hand... is
there a contrib package out there that may help me? Should I program the
function on my own?

  Thanks in advance,

           Fabrizio Ferri Benedetti



-- 
"La mayor sabidur??a que existe es conocerse a uno mismo" - Galileo
Galilei

1) I am not sure why you want the biased version but here are two ways
to do this. Also if you have large enough sample size, it would not
matter much.

	my.var1 <- function(x) mean( (x - mean(x))^2 )
	my.var2 <- function(x) mean( x^2 - mean(x)^2 )

2) Look at quantiles function. Your equation might correspond to one of
the types given there.

Regards, Adai



On Fri, 2005-09-30 at 10:46 +0200, Fabrizio Ferri Benedetti
wrote:
>   Hello everybody (I'm new here)
> 
>   I am looking for 2 specific functions. The first one is a variance
> formula that does not divide by n-1 (I don't want the unbiased formula,
> that's it). In case it does not exist, how do I modify the default
> function code? Yes, a workaround would be to multiply the result by n-1
> and then divide it by n, but I'd like to use a separate function, if
> possible.
> 
>   The other function I'm looking for it's a median formula like
this one:
> 
> http://www.faberitius.com/prototipo/gfx/clip_image002_0000.gif
> 
>   Where "Li" is the exact lower limit of the critical interval,
"nb" is
> the number of cases that lie below the critical interval, "nd"
the number
> of cases inside the mentioned interval, and "I" is the range of
the
> interval itself. As you can see this formula calculates the median in
> categorical structured data.
> 
>   I couldn't find any functions that match the results done by hand...
is
> there a contrib package out there that may help me? Should I program the
> function on my own?
> 
>   Thanks in advance,
> 
>            Fabrizio Ferri Benedetti
> 
> 
>