Re: Special Excel-meaning of median.

["Andreas J. Guelzow" <aguelzow math concordia ab ca>]

Morten Welinder wrote:

> Oh, boy.

yeah...

> In *my* book, the median (singular) and any other fractile is an
> observation in the data set.  More precisely, it is the largest
> of the medians (plural).  If the destribution function is called
> F, the medians are the members of the set
>
>     { x | lim [xm -> x-] F(xm) <= p <= F(x) }
>
> where, p=1/2.  (That would be the closed interval between the two
> points that Excel average over in the discrete case.)
>
> But we're shooting for Excel compatibility, so all that is not
> very important.

good thing. (In fact this `Excel' definition is the same as the one used 
by MuPad and ( I am pretty sure) Mathematica, as well as the definition 
used in most North American intro stats books. Of course which one is 
`correct' I would rather not argue.)

> > In any case, it would be nice to use this function in analysis-tools.c. 
> > That's why I am wondering whether this function shouldn't be in 
> > math-func.h/c with all the other range_* functions rather than in 
> > fn-stats.c?
>
>
> I think it may be time to extract all the range functions (none of
> which are related to R code, btw.) out into a new file, say rangefuncs.c.
> That would make them callable from analysis-tools.c
>
> Morten

Having the range_* functions in math-func.* is fine, analysis-tools.c 
needs some of the other functions in math-func.h anyways. It would just 
be nice to have range_excel_median together with the otehr range_* 
functions.

Andreas

--
Prof. Dr. Andreas J. Guelzow                    
Assoc. Prof of Mathematics
Concordia University College of Alberta
http://www.math.concordia.ab.ca/aguelzow