Hi All,
Thanks for the clarification. Now, perhaps I should use kappa instead.
Since my predictions are in 1 and 2, there are no numeric predictions. To my
surprise, when I applied kappa and auc to the data, their values are highly
correlated, with only an exception when there are perfect predict
Please note that predicted1 and predicted2 are two sets of predictions instead
of predictors. As you can see the predictions with only two levels, 1 is for
hard and 2 for soft. I need to assess which one is more accurate. Hope this is
clear now. Thanks.
Jin
-Original Message-
From: Davi
Hi all,
I used the following code and data to get auc values for two sets of
predictions:
library(caret)
> table(predicted1, trainy)
trainy
hard soft
1 270
2 11 99
> aucRoc(roc(predicted1, trainy))
[1] 0.5
> table(predicted2, trainy)
trainy
hard soft
1
Try this:
> a_tmp<-c("a", "b", "d", "e", "f", "g", 'h')
> b_tmp<-c("a", "c", "e", "g")
> setdiff(b_tmp, a_tmp)
[1] "c"
> setdiff(a_tmp, b_tmp)
[1] "b" "d" "f" "h"
Is this a bug?
Jin
-Original Message-
From: r-help-boun...@r-project.org [mailto:r-help-boun...@r-project.org] On
Behalf Of Gab
Hi Miltinho,
There is a brief review on issues about and a new method for modelling
presence-only data in DOI 10.1007/s10531-007-9270-7. Hope it is useful.
Cheers,
Jin
Jin Li, PhD
Spatial Modeller/
Computational Statistician
Marine & Coastal Environm
Try hclust with daisy in cluster package.
Cheers,
Jin
Jin Li, PhD
Spatial Modeller/
Computational Statistician
Marine & Coastal Environment
Geoscience Australia
Ph: 61 (02) 6249 9899
Fax: 61 (02) 6249 9956
email: [EMAIL PROTECTED]
---
Dear Prof Ripley and Don,
Thank you all for your kind reply.
The UNIX system we have is Sparc Solaris v8 (a 64bit version). The
Precompiled binary distributions of the base system and contributed packages
are only available for Linux, MacOS X and Windows. The only thing available
for UNIX seems S
Thanks to all for your kind suggestions.
After some discussion with our IT staff, I was told the UNIX system we have
is Solaris and installation of R is very time consuming because "Given that
this software is not standard, and given the amount of time required to
compile the software (and potenti
Thank you very much for the explanation and clarification, Patrick. It is a
great help.
Best regards,
Jin
-Original Message-
From: Patrick Burns [mailto:[EMAIL PROTECTED]
Sent: Friday, 25 January 2008 9:48
To: Li Jin
Cc: [EMAIL PROTECTED]; r-help@r-project.org
Subject: Re: [R] Linux, UNI
Hi Peter,
Thank you for the quick answer.
Given that we have UNIX system available here in Geoscience Australia, I am
wondering if UNIX is also faster and with 64 bit build. Perhaps they may be
kind enough to let me have a Linux system.
As I only have experience in using Windows PC, what kind of c
Dear All,
I am currently using R in Windows PC with a 2 GB of RAM. Some pretty large
datasets are expected soon, perhaps in an order of several GB. I am facing a
similar situation like Ralph, either to get a new PC with a bigger RAM or
else. I am just wondering if R is getting faster in other syste
Hi All,
I installed R 2.5.1 recently on a PC (Windows XP Professional 2001) and tried
to install some R packages. It took several minutes and gave me the following
message.
> utils:::menuInstallPkgs()
--- Please select a CRAN mirror for use in this session ---
Error in open.connection(fil
12 matches
Mail list logo