Date:         Tue, 26 Jan 1999 12:33:12 -0500
Reply-To:     "Self, Karsten" <Karsten.Self@SCHWAB.COM>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         "Self, Karsten" <Karsten.Self@SCHWAB.COM>
Subject:      Re: SAS Q: Automatic aesthetic major tick marks on gplot axes
Comments: To: "Hudson, Spencer" <shudson@VIROPHARMA.COM>
Content-Type: text/plain

Conceptual response.
 
Yes, it can be done.
 
Nice is subjective, you (or your client) is the subject.  Define 'nice'.
 
Determine your data range, treatment of outliers (include/exclude),
treatement of origin (or axis origin), and a ballpark idea of what
reasonable rounding factors are.  Probably involves pre-processing your data
with one or more of univariate statistics and/or DATA step, deciding on
parameters, and remembering things to tell your PROC PLOT step.
 
"Easy" is also subjective.
 
--
Karsten M. Self (Karsten.Self@schwab.com)
Trilogy Consulting
 
    What part of "gestalt" don't you understand?
 
WARNING:  All e-mail sent to or from this address will be received by the
Charles Schwab corporate  e-mail system and is subject to archival and
review by someone other than the recipient.
 
This communication does not necessarily reflect the views or position of The
Charles Schwab Corporation, or any of its subsidiaries or their affiliates.
 
 
> ----------
> From:         Hudson, Spencer[SMTP:shudson@VIROPHARMA.COM]
> Reply To:     Hudson, Spencer
> Sent:         Tuesday, January 26, 1999 5:20 AM
> To:   SAS-L@UGA.CC.UGA.EDU
> Subject:      SAS Q: Automatic aesthetic major tick marks on gplot axes
>
> Given a range of data, is there an easy way of automatically determining
> a set of aesthetically pleasing values at which to draw and label major
> tick marks on a gplot. For example, a panel of clinical laboratory tests
> may include serum cholesterol with a range of 73 to 352 for which a
> 'nice' set might be 70 to 360 by 30. The range of values of serum
> potassium has a range of 2.4 to 11.6, for which a 'nice' set might be 2
> to 12 by 2. Can this choice of values be automated? If so, how? Possibly
> a more interesting and difficult question would be: How do we define
> 'nice'? I have looked in both of Cleveland's book and Tufte's first book
> and could not find much guidance, other than the usual advice of not
> using too many ticks and use rounded values.
>
> Any ideas will be appreciated.
> Cheers,
> Spencer Hudson
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> ~
> Spencer Hudson, Ph.D.
> Director, Biostatistics and Clinical Data Management
> ViroPharma Incorporated
> 405 Eagleview Blvd
> Exton, PA 19341
> Telephone:      (610) 458-7300 ext. 154
> FAX:    (610) 458-7380
> Ticker:         VPHM on NASDAQ
> Electronic mail:        shudson@viropharma.com
> Internet:       www.viropharma.com
>