Ric - I use this scaling function:
scaleNums=: 3 : 0
0 1 scaleNums y
:
'low hi'=. ((<./),>./)x
if. 0 *./ . = 11 o. ,y do.
nums=. y-<./,y
nums=. nums%>./,nums NB. scale from 0 to 1
else.
nums=. y-<./10 o. ,y
nums=. nums%>./10 o. ,nums NB. scale from 0 to 1 for complex
end.
nums=. low+nums*hi-low
)
It could be terser but I wrote it quite a while ago.
On Wed, Sep 15, 2010 at 7:47 PM, Sherlock, Ric <[email protected]>wrote:
> Prompted by the recent Hough transform thread, I looked for a "rescale"
> verb in the system library that would take a set of data and linearly
> rescales it to a different range. I didn't find one and wondered if it might
> be a useful addition.
>
> Admittedly it is a pretty simple thing to reimplement but it seems like a
> task that would come up reasonably frequently.
>
> Here is a stab at one. Any ideas/suggestions for improvement?
>
> NB.*rescale v Linearly rescales data to new range
> NB. usage: [targetrange [,targetmin]] rescale data[;min,max]
> NB. y is: numeric array
> NB. Or 2-item boxed list containing
> NB. 0{:: numeric data array
> NB. 1{:: 2-item numeric list: the minimum and maximum possible values
> for data
> NB. x is: optional 1- or 2-item numeric list
> NB. eg: 15 rescale 8 4 9 5 3.5 7;0 10
> rescale=: 3 : 0
> 1 rescale y
> :
> 'data minmax'=. 2{. boxopen y
> 'trange tmin'=. 2{. x
> 'smin smax'=. (<./ , >./) minmax,,data
> data=. data %&(-&smin) smax
> tmin + trange * data
> )
>
> rescale 4 1 9 5 3.5 7 NB. obtains min & max from data
> 0.375 0 1 0.5 0.3125 0.75
> rescale 4 1 9 5 3.5 7;0 10 NB. give min & max possible vals if not
> in data
> 0.4 0.1 0.9 0.5 0.35 0.7
> 15 rescale 4 1 9 5 3.5 7;0 10 NB. "out-of-10" to "out-of-15"
> 6 1.5 13.5 7.5 5.25 10.5
> 15 _5 rescale 4 1 9 5 3.5 7;0 10 NB. "out-of-10" to "from _5 to 10"
> 1 _3.5 8.5 2.5 0.25 5.5
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
--
Devon McCormick, CFA
^me^ at acm.
org is my
preferred e-mail
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
