Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > sci.stat.math > #10900

Re: Q predicting the direction of change due to the change of input

Show all headers | View raw

On Sun, 16 Jul 2023 01:55:21 -0700 (PDT), Cosine <asecant@gmail.com>
wrote:

>Hi:
>
 >  Given a function, y = f(a, b, c, d, ...) we want to know if the 
direction of the change of a specified input variable would result in 
a particular direction of the change of y. Say, setting the value of a
to be higher than a specified value (a_spec) would result in an 
increase of the output y, and a lower value would result in a lower y.
>
>  How do we setup a test for this purpose?
>

If each of (b, c, d, ...) have only one possible value, then you would
only have to be concerned with the possible values of a and y.  
Graph what you think is conceivable, mark off what is interesting, 
and WHAT TEST should become inevitable. 

Discrete values? specifically a linear relation? only a linear 
relation?  That's a starting point for whatever is realistic for
more realistic data where there are other values of (b, ...).

Are all the values 'designed' or is this largely naturalistic
observation, where you only can/will interfere with a?  Tell us more. 

 - I doubt that anyone else will be more concrete in suggestions
unless you offer specific details. 

-- 
Rich Ulrich 

Back to sci.stat.math | Previous | Next — Previous in thread | Find similar

Thread

Q predicting the direction of change due to the change of input Cosine <asecant@gmail.com> - 2023-07-16 01:55 -0700
  Re: Q predicting the direction of change due to the change of input Rich Ulrich <rich.ulrich@comcast.net> - 2023-07-16 17:16 -0400

csiph-web