s-news
[Top] [All Lists]

Re: Identifying the lines in a km plot

To: s-news@lists.biostat.wustl.edu
Subject: Re: Identifying the lines in a km plot
From: "Eric Lim" <ericlim2@hotmail.com>
Date: Thu, 07 Aug 2003 09:35:34 +0100
Dear all,

Thank you very much for the help you have provided on this matter. It has certainly made things much clearer for me. I have enclosed a very clear email from Richard Pugh that I found most useful, but was not posted on the S-news site.

Just to provide some feedback:

Professor Ripley: Thank you for your suggestion, unfortunately when I used

fit$strata

All I got back was NULL, I?m wondering if I used it correctly?

Professor Harrell: Thank you. I?ve just discovered the design library after buying your book and have tried it out, and it does work, although I am still grappling with a number of issues regarding customising the labels and size of the output for the n.risk functions. It is a very useful command.


Sincerely,


Eric Lim
Cambridge, UK.



From: "Richard Pugh" <rpugh@mango-solutions.com>
Reply-To: <Rpugh@mango-solutions.com>
To: "'Eric Lim'" <ericlim2@hotmail.com>
Subject: RE: [S] Identifying the lines in a km plot
Date: Tue, 5 Aug 2003 15:10:54 +0100

Hi Eric,

The quick answer is "yes", S-PLUS will use them in numeric order in this
instance.  However, that's not the end of the answer ...

If your "group" variable is not a factor, it will be converted into one.
S will construct the factor levels in numeric / alphabetical order ...
In your case, the numbers 1 to 4 are converted to factor levels "1" to
"4" and given the line types in order as you supply them (1, 4, 6 then
8).

In your example of "Red", "Black", "Blue" and "Green", S will construct
the factor in alphabetical order ...

> levels(factor(sample(c("Red", "Black", "Blue", "Green"), 100, T)))
[1] "Black" "Blue"  "Green" "Red"

... so care must be taken!  Tricky problem to label "red" with red lines
etc.  I would possibly do it using ...

pie(rep(1, 16), col=1:16)

.. then set up a vector for matching:-

ColNum  Col
1               Black
2               Blue
..              ...

That's one idea .. there may be an easier way to do that ...

Hope this helps,
Rich.

Mango Solutions
Tel : (01628) 418134
Mob : (07967) 808091

_________________________________________________________________
Tired of 56k? Get a FREE BT Broadband connection http://www.msn.co.uk/specials/btbroadband


<Prev in Thread] Current Thread [Next in Thread>