Date: Dec 14, 2012 3:39 PM
Author: Ray Koopman
Subject: Re: One other general question regarding scaling to [0,1].
On Dec 14, 6:04 am, djh <halitsk...@att.net> wrote:
> When comparing one or more (x,y) plots to the reference line segment
> [(0,0),(1,1)] (as you instructed in your post of 12/13 @ 6/27:
> ?The best way to see the difference between the plots is to take cols
> 2 & 3 as x & y coordinates, then plot the points along with a line
> from (0,0) to (1,1).?
> do all x and y values have to be scaled to [0,1] before plotting the
> (x,y) pairs?
> If so, what is the ?correct? (or ?best? or ?standard?) way to scale
> the numbers for:
> a) one pair of (x,y) columns?
> b) more than one pair of (x,y) columns ?
> Thanks for considering this additional question.
In this case all the values were naturally in [0,1], so the reference
line was obvious once I'd seen the plot. In general, tho, there is no
standard procedure. You need both some idea of how the curves might
differ and a "sense of the data" to know what rescalings may be
permissible. It's an art, not a science.