date:20130905

I've also never had a problem with both 32 and 64 bit java installed.

Best,
Ista

On Wed, Sep 4, 2013 at 9:50 PM, Joshua Wiley jwiley.ps...@gmail.com wrote:
 Hi Spencer,

 It really is not very hard, and I have never had issue with it:

 http://www.oracle.com/technetwork/java/javase/downloads/jdk7-downloads-1880260.html

 Just download the x86 and x64 versions for your OS and install.  Worst
 case, you need to add the directory to the PATH variable in Windows.

 I do this regularly so I can use/test either version of R.

 Cheers,

 Josh

 P.S. Emacs + ESS allows for different versions of R and it is not too
 difficult to use the 64 or 32 bit version... M-x
 R-version-architecture


 On Wed, Sep 4, 2013 at 6:36 PM, Spencer Graves
 spencer.gra...@structuremonitoring.com wrote:
 On 9/4/2013 6:09 PM, Ista Zahn wrote:

 Hi Spencer,

 Why don't you want to install 64bit Java?



   That may be a reasonable approach.


   I may have Java confused with something else, but I remember hearing
 that it was difficult or unwise to try to install both 32- and 64-bit
 versions of something like Java or Java Script on the same Windows operating
 system.  If I need to uninstall 32-bit Java to install 64-bit, who knows
 what else I could break.  I'm a statistician, not an information
 technologist:  If I spend more time playing with Java, I'll have less time
 for other things I want to do.


   Thanks for the reply.
   Spencer


 On Wed, Sep 4, 2013 at 6:12 PM, Spencer Graves
 spencer.gra...@structuremonitoring.com wrote:

 Hello, All:


 What do you recommend for reading a Google Spreadsheet into R? I didn't
 find
 anything useful using library(sos); findFn('google spreadsheet').


 I can solve the problem by downloading the file either as *.ods or *.xlsx
 format, then opening it and saving it as *.xls, then using
 read.xls{gdata}.


 Alternatives I haven't tried use read.xlsx{xlsx} and
 readWorksheetFromFile{XLConnect} with 32-bit R. Neither of these work for
 me
 with 64-bit R, because they can't find an appropriate rJava on my
 computer;
 see below. (I've been using 64-bit R with Emacs, so switching to 32-bit R
 is
 not completely trivial.) Similarly, read.gnumeric.sheet{gnumeric}
 requires
 the external program, ssconvert, which seems not to be available on my
 computer or installed for 64-bit R.


 What do you suggest? Avoid 64-bit R unless I really need it? That seems
 to
 be the message I'm getting from this. (The writeFindFn2xls{sos} also
 works
 in 32-bit R but fails in 64-bit apparently for the same reason.)


 Thanks,
 Spencer


 library(xlsx)

 Loading required package: xlsxjars
 Loading required package: rJava
 Error : .onLoad failed in loadNamespace() for 'rJava', details:
 call: fun(libname, pkgname)
 error: No CurrentVersion entry in Software/JavaSoft registry! Try
 re-installing Java and make sure R and Java have matching architectures.
 Error: package ‘rJava’ could not be loaded

 library(XLConnect)

 Loading required package: rJava
 Error : .onLoad failed in loadNamespace() for 'rJava', details:
 call: fun(libname, pkgname)
 error: No CurrentVersion entry in Software/JavaSoft registry! Try
 re-installing Java and make sure R and Java have matching architectures.
 Error: package ‘rJava’ could not be loaded

 sessionInfo()

 R version 3.0.1 (2013-05-16)
 Platform: x86_64-w64-mingw32/x64 (64-bit)

 locale:
 [1] LC_COLLATE=English_United States.1252
 [2] LC_CTYPE=English_United States.1252
 [3] LC_MONETARY=English_United States.1252
 [4] LC_NUMERIC=C
 [5] LC_TIME=English_United States.1252

 attached base packages:
 [1] stats graphics grDevices utils datasets methods base

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide
 http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.


 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.



 --
 Joshua Wiley
 Ph.D. Student, Health Psychology
 University of California, Los Angeles
 http://joshuawiley.com/
 Senior Analyst - Elkhart Group Ltd.
 http://elkhartgroup.com

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] xyplot and lwd

2013-09-05 Thread David Winsemius



On Sep 5, 2013, at 2:54 AM, Daniel Hornung wrote:


On Wed, Sep 4, 2013 at 1:45 PM, Daniel Hornung

daniel.horn...@ds.mpg.dewrote:

Hello,

can it be that xyplot does not support the lwd argument?


The lattice plotting system uses the grid plotting engine and does  
accepts some base  par-type arguments but not all. You may need to  
read more about lattice and grid:


?lattice
?trellis.par.set
require(grid)
?gpar

--
David.


At least here, the following still shows thin lines, as opposed to  
the

regular
plot command:

xyplot(Sepal.Length ~ Sepal.Width, data = iris, pch=4, lwd=4)


On Thursday, September 05, 2013 00:33:32 Bert Gunter wrote:
You should get no lines at all, as you have not specified that  
lines be

drawn. Use the type argument to do so.

xyplot(rnorm(5) ~1:5,pch=4)  ## points only
xyplot(rnorm(5) ~1:5,pch=4,type=b,lwd=4) ## points with thick lines

read ?panel.xyplot carefully (the default panel function for  
xyplot) for

details

Cheers,
Bert


Hello Bert,

no, maybe I expressed myself ambiguously: I was referring to the line
thickness of the symbols, not a line between symbols:

xyplot(rnorm(5) ~ 1:5, pch=4, lwd=4)
versus
plot(rnorm(5), 1:5, pch=4, lwd=4)

I would like the points (4(x) in this case) to have thicker lines.

Cheers,
Daniel



David Winsemius, MD
Alameda, CA, USA

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] xyplot and lwd

2013-09-05 Thread Daniel Hornung

On Thursday, September 05, 2013 13:40:00 David Winsemius wrote:
  can it be that xyplot does not support the lwd argument?
 
 The lattice plotting system uses the grid plotting engine and does
 accepts some base  par-type arguments but not all. You may need to
 read more about lattice and grid:
 
 ?lattice
 ?trellis.par.set
 require(grid)
 ?gpar

Thanks for the hint, I will look further into this direction.

Daniel

-- 
Max-Planck-Institute for Dynamics and Self-Organization
Laboratory for Fluid Dynamics, Pattern Formation and Biocomplexity
Biomedical Physics Group

Am Fassberg 17
D-37077 Goettingen

(+49) 551 5176 373


signature.asc
Description: This is a digitally signed message part.
__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Attribute Length Error when Trying plm Regression

2013-09-05 Thread Millo Giovanni

Dear Laura,

as Arun said it is difficult to help w/o a reproducible example. However
this is most likely to be an indexing problem, as he suggests; the
output of traceback() is far from useless here, because it shows that
the problem occurs in the data transformation step. The latter, which is
by default the within transf., wants (correct indexing and-) time
variation in the data. Hence I suggest you 
- better check the data, especially indices, possibly with str() to see
if they have the right type etc.
then do simpler things, complicating step by step until you spot the
critical one:
- try lm(yourformula, yourdata) to see whether there are data problems
w.r.t. OLS (unlikely, but would spot string variables or other common
pitfalls)
- try pdata.frame() using the individual and time index, to spot if the
indices are somehow inappropriate
- try plm(... , model=pooling), which does not transform the data, or
model=random, which allows for time-invariants; both are more tolerant
than within

This is most likely either bad indices or bad data formats/NAs/empty
groups...

Best wishes,
Giovanni

Giovanni Millo, PhD
Research Dept.,
Assicurazioni Generali SpA
Via Machiavelli 3,
34132 Trieste (Italy)
tel. +39 040 671184
fax  +39 040 671160


 Original thread --

Message: 48
Date: Wed, 4 Sep 2013 14:14:23 -0700 (PDT)
From: arun smartpink...@yahoo.com
To: lross8 lro...@kent.edu
Cc: R help r-help@r-project.org
Subject: Re: [R] Attribute Length Error when Trying plm Regression
Message-ID:
1378329263.73618.yahoomail...@web142604.mail.bf1.yahoo.com
Content-Type: text/plain; charset=iso-8859-1

HI,
It is better to provide a reproducible example using ?dput().
you can also check in this link.

http://r.789695.n4.nabble.com/names-attribute-must-be-the-same-length-as
-the-vector-td4503946.html

library(plm)
#Using the example from ?plm()
?data(Produc, package = plm)
?zz - plm(log(gsp) ~ log(pcap) + log(pc) + log(emp) + unemp, data =
Produc, index = c(state,year))

#Suppose, if I use a model like this:
zz1-
plm(gsp~pcap+pc+emp+unemp+water+util,data=Produc,index=c(gsp,year))
#Error in names(y) - namesy : 
?# 'names' attribute [816] must be the same length as the vector [0]

In your model statement, 

fixed - plm (h ~ o + m + a, data=drugsXX, index=c(h,year),
model=within) 


A.K.


- Original Message -
From: lross8 lro...@kent.edu
To: r-help@r-project.org
Cc: 
Sent: Wednesday, September 4, 2013 3:22 PM
Subject: [R] Attribute Length Error when Trying plm Regression

Hello,

I am trying to run a fixed effects panel regression on data containing 5
columns and 1,494 rows. 

I read the data in as follows:

drugsXX-read.csv(file=C:\\Folder\\vX.X\\Drugs\\drugsXX_panel.csv,
head=TRUE, sep=,)

Verified it read in correctly and had a good data.frame:
dim(drugsXX)
[1] 1494? ? 5
drugs XX
produce expected data with correct column names

The issue is, when I go to run the plm using:
fixed - plm (h ~ o + m + a, data=drugsXX, index=c(h,year),
model=within) 

I get this error:
Error in names(y) - namesy : 
? 'names' attribute [996] must be the same length as the vector [0]

I know the data recognizes that I have 5 columns. I also know that
there's
nothing wrong with row 996 (I even want back and checked for hidden
characters in the original .csv file).

traceback() was useless:
4: pmodel.response.pFormula(formula, data, model = model, effect =
effect, 
? ? ?  theta = theta)
3: pmodel.response(formula, data, model = model, effect = effect, 
? ? ?  theta = theta)
2: plm.fit(formula, data, model, effect, random.method, inst.method)
1: plm(h ~ o + m + a, data = drugsXX, index = c(h, 
? ? ?  year), model = within)

What explicit steps can I follow to get my panel regression to run? 

Thank you, 
Laura



--
View this message in context:
http://r.789695.n4.nabble.com/Attribute-Length-Error-when-Trying-plm-Reg
ression-tp4675384.html
Sent from the R help mailing list archive at Nabble.com.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

- end original thread --

 
Ai sensi del D.Lgs. 196/2003 si precisa che le informazi...{{dropped:12}}

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

[R] Problem with converting F to FALSE

2013-09-05 Thread Venkata Kirankumar

Hi,
I have a peculier problem in R-Project that is when my CSV file have one
column with all values as 'F' the R-Project converting this 'F' to FALSE.
Can some one please suggest how to stop this convertion. Because I want to
use 'F' in my calculations and show it in screen. for example my data is
like

sex  group
F   1
F   2
F   3

but when I use read.csv and load the csv file data is converting it to

sex  group
FALSE   1
FALSE   2
FALSE   3
but i want it as source data like

sex group
F  1
F  2
F  3


Thanks in advance,
D V Kiran Kumar

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Problem with converting F to FALSE

2013-09-05 Thread philippe massicotte

Look at colClasses in ?read.csv

 Date: Thu, 5 Sep 2013 18:14:49 +0530
 From: kiran4u2...@gmail.com
 To: r-help@r-project.org
 Subject: [R] Problem with converting F to FALSE

 Hi,
 I have a peculier problem in R-Project that is when my CSV file have one
 column with all values as 'F' the R-Project converting this 'F' to FALSE.
 Can some one please suggest how to stop this convertion. Because I want to
 use 'F' in my calculations and show it in screen. for example my data is
 like

 sex  group
 F   1
 F   2
 F   3

 but when I use read.csv and load the csv file data is converting it to

 sex  group
 FALSE   1
 FALSE   2
 FALSE   3
 but i want it as source data like

 sex group
 F  1
 F  2
 F  3

 Thanks in advance,
 D V Kiran Kumar

   [[alternative HTML version deleted]]

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Problem with converting F to FALSE

Hi,
Try:
dat1-read.table(text=
sex  group
F   1
F   2
F   3
,sep=,header=TRUE,colClasses=c(character,numeric))
 dat1
#  sex group
#1   F 1
#2   F 2
#3   F 3


#if you are using read.csv()
 
dat2-read.csv(new1.csv,sep=,header=TRUE,colClasses=c(character,numeric))
 dat2
#  sex group
#1   F 1
#2   F 2
#3   F 3


#Suppose you want to convert after reading it:
 dat2New-read.csv(new1.csv,sep=,header=TRUE)
 dat2New[,1]-substr(dat2New[,1],1,1)
 dat2New
#  sex group
#1   F 1
#2   F 2
#3   F 3
A.K.


Hi, 
I have a peculier problem in R-Project that is when my CSV file have one 
column with all values as 'F' the R-Project converting this 'F' to FALSE. 
Can some one please suggest how to stop this convertion. Because I want to 
use 'F' in my calculations and show it in screen. for example my data is 
like 

sex  group 
F       1 
F       2 
F       3 

but when I use read.csv and load the csv file data is converting it to 

sex          group 
FALSE       1 
FALSE       2 
FALSE       3 
but i want it as source data like 

sex group 
F      1 
F      2 
F      3 


Thanks in advance, 
D V Kiran Kumar

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Problem with converting F to FALSE

2013-09-05 Thread Joshua Wiley

Hi,

You can either manually specify colClasses or the asis argument.  See
?read.csv for more details.

If you just had those two columns, something like:

 read.table(header = TRUE, text = 
 sex group
 F 1
 T 2
 , colClasses = c(character, integer))

Cheers,

Josh


read.csv(file.csv, colClasses = c(character, integer))




On Thu, Sep 5, 2013 at 5:44 AM, Venkata Kirankumar
kiran4u2...@gmail.com wrote:
 Hi,
 I have a peculier problem in R-Project that is when my CSV file have one
 column with all values as 'F' the R-Project converting this 'F' to FALSE.
 Can some one please suggest how to stop this convertion. Because I want to
 use 'F' in my calculations and show it in screen. for example my data is
 like

 sex  group
 F   1
 F   2
 F   3

 but when I use read.csv and load the csv file data is converting it to

 sex  group
 FALSE   1
 FALSE   2
 FALSE   3
 but i want it as source data like

 sex group
 F  1
 F  2
 F  3


 Thanks in advance,
 D V Kiran Kumar

 [[alternative HTML version deleted]]

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.



-- 
Joshua Wiley
Ph.D. Student, Health Psychology
University of California, Los Angeles
http://joshuawiley.com/
Senior Analyst - Elkhart Group Ltd.
http://elkhartgroup.com

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

[R] Y-axis labels as decimal numbers

2013-09-05 Thread mohan . radhakrishnan

Hi,
 I am able to create a graph with this code but the decimal 
numbers are not plotted accurately because the ylim values are not set 
properly. x-axis is proper.

How do I accurately set the ylim for duration.1 column ?

Thanks,
Mohan

set1$duration- as.POSIXct(paste('2013-08-24', set1$duration))
plot(set1$duration,set1$duration.1,type=b,col = blue,  ylab=, xaxt = 
'n', xlab=,las=2,lwd=2.5, lty=1,cex.axis=2.5)
# now plot you times
axis(1, at = set1$duration, labels = set1$duration, las = 2,cex.axis=2.5)

   duration duration.1
2  16:03:41   0.05
3  17:03:41   0.27
4  18:03:43   1.22
5  19:03:45   1.51
6  20:03:47   1.27
7  21:03:48   1.15
8  22:03:50   1.22
9  23:03:52   1.27
10 00:03:54   1.27
11 01:03:55   1.22
12 02:03:57   1.26
13 03:03:59   1.57
14 04:04:01   1.31
15 05:04:03   1.24


This e-Mail may contain proprietary and confidential information and is sent 
for the intended recipient(s) only.  If by an addressing or transmission error 
this mail has been misdirected to you, you are requested to delete this mail 
immediately. You are also hereby notified that any use, any form of 
reproduction, dissemination, copying, disclosure, modification, distribution 
and/or publication of this e-mail message, contents or its attachment other 
than by its intended recipient/s is strictly prohibited.

Visit us at http://www.polarisFT.com

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

[R] ggplot2: connecting medians of boxes using facet_wrap. Changing median marker.

2013-09-05 Thread Anna Zakrisson Braeunlich

# Dear all,

# Thank you for taking your time.

# What I would like to do:

# (Run the code below to see what I am referring to)

# I want lines connecting the medians of of the boxes. I do not want a 
function, just a simple,
# straight connection between points. I also would like the lines to be 
different: lty=c(1:3)

# Furthermore, I wish to change the line/dot marking the medians to be 
pch=c(1:3). It seems not to be so simple when using facet_wrap? I have 
obviously missed something obvious.

# Finally, I would like the boxes to be filled white and to have the legend 
reflecting this.

# I know that this was many questions, I apologize if some may seem basic. It 
is just that I am
# jumping packages the whole time and sometimes par() adjustments work and 
sometimes not.
# I have searched alot for answers, but as a ggplot2 beginner, I have failed to 
find a solution
# to my problems above.

# I would like to thank the programmers for a great package. The layer 
principle is much easier to work with.



# Some dummy data:
mydata- data.frame(week = factor(rep(c(19, 21, 23, 25, 27, 29, 
31, 33,
   35, 37, 39), each = 45*3)), 
#week
station = factor(rep(c(BY31, B1, H2, H3, H4,
   H5, H6, H7, H8), each = 
15)), #station
organism = factor(rep(c(zpl, ses, cy), each = 5)), 
#organism
var1 = rnorm(1485, mean = rep(c(0, 3, 15), each = 40),
 sd = rep(c(1, 3, 6), each = 20)))

p - ggplot(mydata, aes(x = week, y = var1, fill = organism)) +
  geom_boxplot() +
  facet_wrap(~ station, ncol = 3) +
  theme_bw() +
  theme(strip.background = element_blank())+
  ylab(var1)+
  xlab(week) +
  geom_smooth(aes(group = 1), method=lm, se = F) +  #Here is my problem.
  theme(strip.text.x = element_text(size = 12, colour=black, family=serif, 
angle=00)) +
  theme(axis.text.x = element_text(size = 12, colour=black, family=serif, 
angle=90)) +
  theme(axis.text.y = element_text(size = 12, colour=black, family=serif, 
angle=00)) +
  geom_hline(yintercept=0, linetype=3) #draws dotted line at 0

p

  # method=lm is definately wrong,
  # I just added it to be a ble to draw some lines at all.
  # I also suspect geom_smooth to be wrong.
### TO CLARIFY: I want the medians connected within each level. cy medians 
connected and
  # and ses medians connected and zpl connected. Not cy-ses-zpl, but that is 
perhaps obvious.
### Thank you for your time!
  # with kind regards
  # A. Zakrisson



Anna Zakrisson Braeunlich
PhD student

Department of Ecology, Environment and Plant Sciences
Stockholm University
Svante Arrheniusv. 21A
SE-106 91 Stockholm
Sweden/Sverige

Lives in Berlin.
For paper mail:
Katzbachstr. 21
D-10965, Berlin - Kreuzberg
Germany/Deutschland

E-mail: anna.zakris...@su.se
Tel work: +49-(0)3091541281
Mobile: +49-(0)15777374888
LinkedIn: http://se.linkedin.com/pub/anna-zakrisson-braeunlich/33/5a2/51b

º`. .  `. . `. . º`. .  `. . `. .º`. .  `. . 
`. .º

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

[R] Using qmap module

2013-09-05 Thread Jaya Pudashine

Dear all,
Is there anybody who is using qmap module for bias correction of RCM data.
I have few queries about that. Please let me know if anybody is using
qmap module in R.


Thanking in Advance



-- 
Jaya Pudashine
M.Eng. Water Engineering and Management
Asian Institute of Technology, Thailand
Contact : +66-920598298

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Poly Correlations

Hi Michael,

See comments in line.

On Wed, Sep 4, 2013 at 10:18 PM, Michael Hacker mhac...@nycap.rr.com wrote:
 Dear Colleagues,



 I'm working on a Delphi study comparing perceptions of high school
 technology teachers and university engineering educators about the
 importance of concepts about engineering for HS students to learn as part of
 their fundamental education. I'm actually doing this as part of my Ph.D.

 The survey items (n=37) are categorized into five scales: design, human
 values, modeling, resources, and systems thinking. I'm seeking to determine
 the reliability of these scales and of the overall survey instrument. Since
 I'm working with ordinal data, Chronbach's Alpha probably isn't the best
 statistical tool to use.



 I've literally spent several days learning my way around R-project but am
 struggling with procedures and interpretations.



 I'm aware that there is now a plug-in for R for SPSS that can be downloaded
 ( http://www-01.ibm.com/support/docview.wss?uid=swg21477550
 http://www-01.ibm.com/support/docview.wss?uid=swg21477550 and
 http://gruener.userpage.fu-berlin.de/Essentials%20for%20R%20Installation%20
 Instructions_21.pdf
 http://gruener.userpage.fu-berlin.de/Essentials%20for%20R%20Installation%20I
 nstructions_21.pdf). Just learned that today and I downloaded
 PolyCorrelations.zip from
 https://www.ibm.com/developerworks/community/files/app?lang=en#/file/9f47f9a
 0-7793-4ad5-8bb7-d3fd1a028e44



I would ditch the SPSS/R integration and just run R from RCommander.
You don't need PollyCorrelations.zip or SPSS for this, and trying to
get the R and SPSS talking to each other is just another level of
complication that you don't need.


 I've gotten as far as loading Rcmdr and running some analyses - (Statistics,
 dimensional analysis, scale reliability) and I've generated this output:



 Reliability deleting each item in turn:

 Alpha   Std.Alpha   r(item, total)

 design  0.84450.8490 0.7629

 humanvalues   0.85260.8541 0.7170

 modeling  0.85110.8546 0.7271

 resources0.87120.8757 0.6328

 systems   0.84610.8498 0.7488



 I now would sincerely appreciate some help. At the age of 70, never having
 studied programming, the meaning of these statistics is not apparent.

Understanding these statistics has nothing to do with studying
programming. You need to study statistics!


 For example, I'm not clear if either of these three statistics are Ordinal
 Alpha. Since I'm working with Likert scale items, my advisor suggested that
 I seek an alternative to Chronbach's Alpha to determine reliability.

Since we have no idea how you calculated these statistics there is no
way for us to answer this question.




 So far, here are the steps I have taken:

 I've searched the FAQs

 Searched specifically for answers on the Web

 Played with the software for hours

 Read the accompanying documentation.

 Downloaded and installed Rcmdr

 Downloaded and installed PolyCorrelations.



 I tried running PolyCorrelations  but I get a message that states that this
 requires the Polychor and Gclus libraries. I tried to install them into the
 R console, but no luck.

What does no luck mean?




 I'd also be pleased to work with someone-on-one on a consulting basis if
 someone has the time and inclination.  Hoping to find an individual who
 knows SPSS and R.

Appendix B of http://pareonline.net/pdf/v17n3.pdf shows how to
calculate reliability from ordinal data using R.

Best,
Ista





 Thanks very sincerely for considering this request.



 Michael





 

 END OF MESSAGE

 

 Michael Hacker, Co-Director

 Hofstra University Center for STEM Education Research

 Ph: 518-724-6437

 Cell: 518-229-7300

 Fax: 518-434-6783

 URL: www.Hofstra.edu/CSR




 [[alternative HTML version deleted]]

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Problem with converting F to FALSE

2013-09-05 Thread Keith Jewell


Depending what you're doing with the data, you might want
colClasses=c(factor,numeric)

On 05/09/2013 13:58, Joshua Wiley wrote:

Hi,

You can either manually specify colClasses or the asis argument.  See
?read.csv for more details.

If you just had those two columns, something like:

  read.table(header = TRUE, text = 
  sex group
  F 1
  T 2
  , colClasses = c(character, integer))

Cheers,

Josh


read.csv(file.csv, colClasses = c(character, integer))




On Thu, Sep 5, 2013 at 5:44 AM, Venkata Kirankumar
kiran4u2...@gmail.com  wrote:

Hi,
I have a peculier problem in R-Project that is when my CSV file have one
column with all values as 'F' the R-Project converting this 'F' to FALSE.
Can some one please suggest how to stop this convertion. Because I want to
use 'F' in my calculations and show it in screen. for example my data is
like

sex  group
F   1
F   2
F   3

but when I use read.csv and load the csv file data is converting it to

sex  group
FALSE   1
FALSE   2
FALSE   3
but i want it as source data like

sex group
F  1
F  2
F  3


Thanks in advance,
D V Kiran Kumar


__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

[R] plot densities outside axis

2013-09-05 Thread Dustin Fife

I've been working on a way to visualize a spearman correlation. That seemed
pretty simple:

 generate skewed data
x = rnorm(100)^2
y = .6*x + rnorm(100, 0, sqrt(1-.6^2))

plot(x,y)   regular plot

plot(rank(x),rank(y), xaxt=n, yaxt=n)  ### spearman-like plot

 make axis labels
axis(1, at=quantile(rank(x)), labels=round(quantile(x), digits=2))
axis(2, at=quantile(rank(y)), labels=round(quantile(y), digits=2))

However, transforming the data into ranks eliminates any information we
have about the distributions of the data. My solution to this problem is to
plot the densities outside the x/y axis with the mode of the distribution
pointing away from the plot. I've seen plots like this in textbooks, but
can't think of a way to do this in R.

Any ideas?

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Histogram

2013-09-05 Thread David Carlson

I wasn't suggesting that much detail, but I think the addition
of one sentence in the last paragraph of the Details section
would make it the meaning of the number is a suggestion only
clearer. 

These functions provide a suggested number of bins that may be
modified to produce 'round' breakpoints covering the range of
the values in x.

Added just before the last sentence, Alternatively, . . .

Also pretty() could be added to the See Also section.


David Carlson

-Original Message-
From: Duncan Murdoch [mailto:murdoch.dun...@gmail.com] 
Sent: Wednesday, September 4, 2013 7:00 PM
To: dcarl...@tamu.edu
Cc: 'philippe massicotte'; 'Rui Barradas';
'r-help@R-project.org'
Subject: Re: [R] Histogram

On 13-09-04 4:44 PM, David Carlson wrote: Good question. It
turns out 
that the manual page does not tell
  the whole story.

Do you really think the manual page would be improved if it went
into as 
much detail as you give below?  It does say clearly that breaks
is a 
suggestion only.  I don't think it would be clearer if it
explained 
exactly how the suggestion is used. It would just be more
complicated, 
and less likely to be read.

Duncan Murdoch


  Looking at the source code for hist.default,
  the function starts with the number of breaks suggested by
  nclass.Sturges(), but then this number (or any other number
of
  breaks that you specify) is passed to pretty() along with the
  maximum and the minimum values of the data (ie range(data))
to
  create pretty break intervals. In your example,
  nclass.Sturges() always recommends 8 breaks, but the number
of
  the breaks changes based on the minimum and maximum values.
So
  the only way to get exactly the number of breaks you want is
to
  specify the break intervals yourself.
 
  David Carlson
 
 
  -Original Message-
  From: r-help-boun...@r-project.org
  [mailto:r-help-boun...@r-project.org] On Behalf Of philippe
  massicotte
  Sent: Wednesday, September 4, 2013 3:02 PM
  To: Rui Barradas
  Cc: r-help@R-project.org
  Subject: Re: [R] Histogram
 
  Thank you everyone.
  Try executing this:
  replicate(100, length(hist(rnorm(100), nclass = 10)$counts))
  I'm still not sure why the number of bins (classes) is not
  consistent.
  Thank in advance.
 
  Date: Wed, 4 Sep 2013 20:27:36 +0100
  From: ruipbarra...@sapo.pt
  To: pmassico...@hotmail.com
  CC: r-help@r-project.org
  Subject: Re: [R] Histogram
 
  Hello,
 
  See the arguments 'right' and 'include.lowest' of ?hist.
  To give what you want, try instead
 
  h1 - hist(1:10, 10)  # counts are 2, 1, 1, ...
  h2 - hist(1:10, breaks = 0:10)  # all counts are 1
 
 
  and see the difference between h1 and h2, components
'breaks'
  and 'counts'.
 
  Hope this helps,
 
  Rui Barradas
 
  Em 04-09-2013 19:34, philippe massicotte escreveu:
  Hi everyone.
  I'm currently translating some Matlab code into R. However,
  I realized that the hsit function produce different results
in
  both languages.
  in Matlab, hist(1:10, 10) will produce 10 bins with a count
  of 1 in each, but in R it will produce 9 classes with count
of
  2,1,1,1,1,1,1,1,1.
  I'm a bit embarrassed to ask such question, but why R is
not
  producing 10 classes as requested?
  Thanks in advance,Phil 
 [[alternative HTML version deleted]]
 
  __
  R-help@r-project.org mailing list
  https://stat.ethz.ch/mailman/listinfo/r-help
  PLEASE do read the posting guide
  http://www.R-project.org/posting-guide.html
  and provide commented, minimal, self-contained,
reproducible
  code.
 
   
   [[alternative HTML version deleted]]
 
  __
  R-help@r-project.org mailing list
  https://stat.ethz.ch/mailman/listinfo/r-help
  PLEASE do read the posting guide
  http://www.R-project.org/posting-guide.html
  and provide commented, minimal, self-contained, reproducible
  code.
 
  __
  R-help@r-project.org mailing list
  https://stat.ethz.ch/mailman/listinfo/r-help
  PLEASE do read the posting guide 
http://www.R-project.org/posting-guide.html
  and provide commented, minimal, self-contained, reproducible
code.
 

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] xyplot and lwd

Daniel:

I wondered if that might be what you meant ...

To amplify a bit on David's response, the answer is that you do **not**
have separate control over the line width of characters -- lwd controls the
width of lines in a graph (exactly as it does in base graphics! ), so you
misunderstood the lwd parameter in the first place. The cex parameter
controls the overall size of plotting characters (and text), so that
incidentally affects the thickness of lines in character rendering. To get
different line thicknesses without changing the overall size, you need to
use different characters, fonts (e.g. bold), or font families, for which
details can be found on the gpar man page, as David said. Note that some of
this may also be device and system dependent,

Cheers,
Bert


On Thu, Sep 5, 2013 at 4:52 AM, Daniel Hornung daniel.horn...@ds.mpg.dewrote:

 On Thursday, September 05, 2013 13:40:00 David Winsemius wrote:
   can it be that xyplot does not support the lwd argument?
 
  The lattice plotting system uses the grid plotting engine and does
  accepts some base  par-type arguments but not all. You may need to
  read more about lattice and grid:
 
  ?lattice
  ?trellis.par.set
  require(grid)
  ?gpar

 Thanks for the hint, I will look further into this direction.

 Daniel

 --
 Max-Planck-Institute for Dynamics and Self-Organization
 Laboratory for Fluid Dynamics, Pattern Formation and Biocomplexity
 Biomedical Physics Group

 Am Fassberg 17
 D-37077 Goettingen

 (+49) 551 5176 373

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide
 http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.




-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] optim evils

2013-09-05 Thread Martin Maechler


 Thanks for all replies.
 The problem occurred in the following context:

 A Gaussian one dimensional mixture (number of constituents, locations, 
 variances all unknown)
 is to be fitted to data (as starting value to or in lieu of mixtools). A 
 likelihood maximization is performed.

Cool.  That is all provided with my  nor1mix  CRAN package  (of which
most parts I have written even before R came to life, i.e., for S):

The relatively new addition to  nor1mix  is the 
norMixMLE() function which uses my smart (almost) unconstrained
parametrization and hence typically works much better, i.e., faster than the EM.

Then, the code uses optim(), currently always with BFGS.

Martin Maechler, ETH Zurich

 I'll try to destill the code so that reproducible failure of L-BFGS-B occurs
 and post it here.


 Michael Meyer


 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Histogram

2013-09-05 Thread Duncan Murdoch


On 05/09/2013 10:17 AM, David Carlson wrote:

I wasn't suggesting that much detail, but I think the addition
of one sentence in the last paragraph of the Details section
would make it the meaning of the number is a suggestion only
clearer.

These functions provide a suggested number of bins that may be
modified to produce 'round' breakpoints covering the range of
the values in x.


I think that's the wrong place for it (since breaks=10 is perfectly 
fine, but is not a function).  I'll change the initial sentence to say:


In the last three cases the number is a suggestion only; the
 breakpoints will be set to \code{\link{pretty}} values.

If people want to know what pretty values are, they can follow the link.

Duncan Murdoch



Added just before the last sentence, Alternatively, . . .

Also pretty() could be added to the See Also section.


David Carlson

-Original Message-
From: Duncan Murdoch [mailto:murdoch.dun...@gmail.com]
Sent: Wednesday, September 4, 2013 7:00 PM
To: dcarl...@tamu.edu
Cc: 'philippe massicotte'; 'Rui Barradas';
'r-help@R-project.org'
Subject: Re: [R] Histogram

On 13-09-04 4:44 PM, David Carlson wrote: Good question. It
turns out
that the manual page does not tell
   the whole story.

Do you really think the manual page would be improved if it went
into as
much detail as you give below?  It does say clearly that breaks
is a
suggestion only.  I don't think it would be clearer if it
explained
exactly how the suggestion is used. It would just be more
complicated,
and less likely to be read.

Duncan Murdoch


   Looking at the source code for hist.default,
   the function starts with the number of breaks suggested by
   nclass.Sturges(), but then this number (or any other number
of
   breaks that you specify) is passed to pretty() along with the
   maximum and the minimum values of the data (ie range(data))
to
   create pretty break intervals. In your example,
   nclass.Sturges() always recommends 8 breaks, but the number
of
   the breaks changes based on the minimum and maximum values.
So
   the only way to get exactly the number of breaks you want is
to
   specify the break intervals yourself.
  
   David Carlson
  
  
   -Original Message-
   From: r-help-boun...@r-project.org
   [mailto:r-help-boun...@r-project.org] On Behalf Of philippe
   massicotte
   Sent: Wednesday, September 4, 2013 3:02 PM
   To: Rui Barradas
   Cc: r-help@R-project.org
   Subject: Re: [R] Histogram
  
   Thank you everyone.
   Try executing this:
   replicate(100, length(hist(rnorm(100), nclass = 10)$counts))
   I'm still not sure why the number of bins (classes) is not
   consistent.
   Thank in advance.
  
   Date: Wed, 4 Sep 2013 20:27:36 +0100
   From: ruipbarra...@sapo.pt
   To: pmassico...@hotmail.com
   CC: r-help@r-project.org
   Subject: Re: [R] Histogram
  
   Hello,
  
   See the arguments 'right' and 'include.lowest' of ?hist.
   To give what you want, try instead
  
   h1 - hist(1:10, 10)  # counts are 2, 1, 1, ...
   h2 - hist(1:10, breaks = 0:10)  # all counts are 1
  
  
   and see the difference between h1 and h2, components
'breaks'
   and 'counts'.
  
   Hope this helps,
  
   Rui Barradas
  
   Em 04-09-2013 19:34, philippe massicotte escreveu:
   Hi everyone.
   I'm currently translating some Matlab code into R. However,
   I realized that the hsit function produce different results
in
   both languages.
   in Matlab, hist(1:10, 10) will produce 10 bins with a count
   of 1 in each, but in R it will produce 9 classes with count
of
   2,1,1,1,1,1,1,1,1.
   I'm a bit embarrassed to ask such question, but why R is
not
   producing 10 classes as requested?
   Thanks in advance,Phil   
[[alternative HTML version deleted]]
  
   __
   R-help@r-project.org mailing list
   https://stat.ethz.ch/mailman/listinfo/r-help
   PLEASE do read the posting guide
   http://www.R-project.org/posting-guide.html
   and provide commented, minimal, self-contained,
reproducible
   code.
  

[[alternative HTML version deleted]]
  
   __
   R-help@r-project.org mailing list
   https://stat.ethz.ch/mailman/listinfo/r-help
   PLEASE do read the posting guide
   http://www.R-project.org/posting-guide.html
   and provide commented, minimal, self-contained, reproducible
   code.
  
   __
   R-help@r-project.org mailing list
   https://stat.ethz.ch/mailman/listinfo/r-help
   PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
   and provide commented, minimal, self-contained, reproducible
code.
  




__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible

Re: [R] ggplot2: connecting medians of boxes using facet_wrap. Changing median marker.

On Thu, Sep 5, 2013 at 11:08 AM, Anna Zakrisson Braeunlich
anna.zakris...@su.se wrote:
 Hi and thank you for the help and for the fast reply!

 A. One thing still is a problem. I have prior my first mail tried to not fill 
 the boxes. The result is a different plot.
 If chosing to do:

 ggplot(mydata, aes(x = week, y = var1))

 instead of

 ggplot(mydata, aes(x = week, y = var1, fill=organism))

 I get the median of all the levels of organisms plotted per week and not the 
 median per organism level per week.
 I would like to have white boxes, but to define each level of organism with a 
 different non-filled symbol that mark the median of each box.

Ooops, sorry I missed that before. You can use

geom_boxplot(mapping=aes(group=interaction(week, organism)))

to achieve this, though it now becomes hard to tell which boxplots
correspond to which organisms.


 B. Can I define the shapes of the geom_point other than the default filled 
 points that I will automatically generate when I define:
 geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) + # 
 I need non-filled symbols

add

scale_shape(solid=FALSE)

or

scale_shape_manual(values = 1:3)

Best,
Ista

 The two versions mentioned in A are:

 ###DATA
 mydata- data.frame(week = factor(rep(c(19, 21, 23, 25, 27, 29, 
 31, 33,
35, 37, 39), each = 45*3)), 
 #week
 station = factor(rep(c(BY31, B1, H2, H3, H4,
H5, H6, H7, H8), each = 
 15)), #station
 organism = factor(rep(c(zpl, ses, cy), each = 5)), 
 #organism
 var1 = rnorm(1485, mean = rep(c(0, 3, 15), each = 40),
  sd = rep(c(1, 3, 6), each = 20)))

 ###VERSION 1
 p - ggplot(mydata, aes(x = week, y = var1)) + #here all organisms are bunted 
 together
   geom_boxplot() +
   facet_wrap(~ station, ncol = 3) +
   theme_bw() +
   theme(strip.background = element_blank())+
   ylab(var1)+
   xlab(week) +
   geom_line(stat=summary, fun.y = median,
 mapping=aes(linetype=organism, group=organism))+ #must add jitter 
 if using this
   geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) + 
 #must be unfilled
   theme(strip.text.x = element_text(size = 12, colour=black, 
 family=serif, angle=00)) +
   theme(axis.text.x = element_text(size = 12, colour=black, family=serif, 
 angle=90)) +
   theme(axis.text.y = element_text(size = 12, colour=black, family=serif, 
 angle=00)) +
   geom_hline(yintercept=0, linetype=3) #draws dotted line at 0

 p

 ###VERSION 2

 p - ggplot(mydata, aes(x = week, y = var1, fill=organism)) + #here the 
 levels in organism are considered
   geom_boxplot() +
   facet_wrap(~ station, ncol = 3) +
   theme_bw() +
   theme(strip.background = element_blank())+
   ylab(var1)+
   xlab(week) +
   geom_line(stat=summary, fun.y = median,
 mapping=aes(linetype=organism, group=organism))+ #must add jitter 
 if using this
   geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) + 
 #must be unfilled
   theme(strip.text.x = element_text(size = 12, colour=black, 
 family=serif, angle=00)) +
   theme(axis.text.x = element_text(size = 12, colour=black, family=serif, 
 angle=90)) +
   theme(axis.text.y = element_text(size = 12, colour=black, family=serif, 
 angle=00)) +
   geom_hline(yintercept=0, linetype=3) #draws dotted line at 0

 p
 #




 Anna Zakrisson Braeunlich
 PhD student

 Department of Ecology, Environment and Plant Sciences
 Stockholm University
 Svante Arrheniusv. 21A
 SE-106 91 Stockholm
 Sweden/Sverige

 Lives in Berlin.
 For paper mail:
 Katzbachstr. 21
 D-10965, Berlin - Kreuzberg
 Germany/Deutschland

 E-mail: anna.zakris...@su.se
 Tel work: +49-(0)3091541281
 Mobile: +49-(0)15777374888
 LinkedIn: http://se.linkedin.com/pub/anna-zakrisson-braeunlich/33/5a2/51b

º`•. . • `•. .• `•. . º`•. . • `•. .• `•. .º`•. . • `•. 
.• `•. .º

 
 From: Ista Zahn [istaz...@gmail.com]
 Sent: 05 September 2013 16:02
 To: Anna Zakrisson Braeunlich
 Cc: r-help@r-project.org
 Subject: Re: [R] ggplot2: connecting medians of boxes using facet_wrap. 
 Changing median marker.

 Hi Anna,

 On Thu, Sep 5, 2013 at 6:13 AM, Anna Zakrisson Braeunlich
 anna.zakris...@su.se wrote:
 # Dear all,

 # Thank you for taking your time.

 # What I would like to do:

 # (Run the code below to see what I am referring to)

 # I want lines connecting the medians of of the boxes. I do not want a 
 function, just a simple,
 # straight connection between points. I also would like the lines to be 
 different: lty=c(1:3)

 geom_line(stat=summary, fun.y = median,
 mapping=aes(linetype=organism,

Re: [R] ggplot2: connecting medians of boxes using facet_wrap. Changing median marker.

Hi Anna,

On Thu, Sep 5, 2013 at 6:13 AM, Anna Zakrisson Braeunlich
anna.zakris...@su.se wrote:
 # Dear all,

 # Thank you for taking your time.

 # What I would like to do:

 # (Run the code below to see what I am referring to)

 # I want lines connecting the medians of of the boxes. I do not want a 
 function, just a simple,
 # straight connection between points. I also would like the lines to be 
 different: lty=c(1:3)

geom_line(stat=summary, fun.y = median,
mapping=aes(linetype=organism, group=organism))


 # Furthermore, I wish to change the line/dot marking the medians to be 
 pch=c(1:3). It seems not to be so simple when using facet_wrap? I have 
 obviously missed something obvious.

geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) +


 # Finally, I would like the boxes to be filled white and to have the legend 
 reflecting this.

Then don't tell ggplot to color them. Change

ggplot(mydata, aes(x = week, y = var1, fill = organism))

to

ggplot(mydata, aes(x = week, y = var1))

 # I know that this was many questions, I apologize if some may seem basic. It 
 is just that I am
 # jumping packages the whole time and sometimes par() adjustments work and 
 sometimes not.
 # I have searched alot for answers, but as a ggplot2 beginner, I have failed 
 to find a solution
 # to my problems above.


Hope this helps!

Best,
Ista


 # I would like to thank the programmers for a great package. The layer 
 principle is much easier to work with.

 

 # Some dummy data:
 mydata- data.frame(week = factor(rep(c(19, 21, 23, 25, 27, 29, 
 31, 33,
35, 37, 39), each = 45*3)), 
 #week
 station = factor(rep(c(BY31, B1, H2, H3, H4,
H5, H6, H7, H8), each = 
 15)), #station
 organism = factor(rep(c(zpl, ses, cy), each = 5)), 
 #organism
 var1 = rnorm(1485, mean = rep(c(0, 3, 15), each = 40),
  sd = rep(c(1, 3, 6), each = 20)))

 p - ggplot(mydata, aes(x = week, y = var1, fill = organism)) +
   geom_boxplot() +
   facet_wrap(~ station, ncol = 3) +
   theme_bw() +
   theme(strip.background = element_blank())+
   ylab(var1)+
   xlab(week) +
   geom_smooth(aes(group = 1), method=lm, se = F) +  #Here is my problem.
   theme(strip.text.x = element_text(size = 12, colour=black, 
 family=serif, angle=00)) +
   theme(axis.text.x = element_text(size = 12, colour=black, family=serif, 
 angle=90)) +
   theme(axis.text.y = element_text(size = 12, colour=black, family=serif, 
 angle=00)) +
   geom_hline(yintercept=0, linetype=3) #draws dotted line at 0

 p

   # method=lm is definately wrong,
   # I just added it to be a ble to draw some lines at all.
   # I also suspect geom_smooth to be wrong.
 ### TO CLARIFY: I want the medians connected within each level. cy medians 
 connected and
   # and ses medians connected and zpl connected. Not cy-ses-zpl, but that is 
 perhaps obvious.
 ### Thank you for your time!
   # with kind regards
   # A. Zakrisson



 Anna Zakrisson Braeunlich
 PhD student

 Department of Ecology, Environment and Plant Sciences
 Stockholm University
 Svante Arrheniusv. 21A
 SE-106 91 Stockholm
 Sweden/Sverige

 Lives in Berlin.
 For paper mail:
 Katzbachstr. 21
 D-10965, Berlin - Kreuzberg
 Germany/Deutschland

 E-mail: anna.zakris...@su.se
 Tel work: +49-(0)3091541281
 Mobile: +49-(0)15777374888
 LinkedIn: http://se.linkedin.com/pub/anna-zakrisson-braeunlich/33/5a2/51b

º`•. . • `•. .• `•. . º`•. . • `•. .• `•. .º`•. . • `•. 
.• `•. .º

 [[alternative HTML version deleted]]


 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.


__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] ggplot2: connecting medians of boxes using facet_wrap. Changing median marker.

2013-09-05 Thread Anna Zakrisson Braeunlich

Hi and thank you for the help and for the fast reply!

A. One thing still is a problem. I have prior my first mail tried to not fill 
the boxes. The result is a different plot.
If chosing to do:

ggplot(mydata, aes(x = week, y = var1))

instead of 

ggplot(mydata, aes(x = week, y = var1, fill=organism))

I get the median of all the levels of organisms plotted per week and not the 
median per organism level per week.
I would like to have white boxes, but to define each level of organism with a 
different non-filled symbol that mark the median of each box.

B. Can I define the shapes of the geom_point other than the default filled 
points that I will automatically generate when I define:
geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) + # I 
need non-filled symbols

The two versions mentioned in A are:

###DATA
mydata- data.frame(week = factor(rep(c(19, 21, 23, 25, 27, 29, 
31, 33,
   35, 37, 39), each = 45*3)), 
#week
station = factor(rep(c(BY31, B1, H2, H3, H4,
   H5, H6, H7, H8), each = 
15)), #station
organism = factor(rep(c(zpl, ses, cy), each = 5)), 
#organism
var1 = rnorm(1485, mean = rep(c(0, 3, 15), each = 40),
 sd = rep(c(1, 3, 6), each = 20)))

###VERSION 1
p - ggplot(mydata, aes(x = week, y = var1)) + #here all organisms are bunted 
together
  geom_boxplot() +
  facet_wrap(~ station, ncol = 3) +
  theme_bw() +
  theme(strip.background = element_blank())+
  ylab(var1)+
  xlab(week) +
  geom_line(stat=summary, fun.y = median,
mapping=aes(linetype=organism, group=organism))+ #must add jitter 
if using this
  geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) + 
#must be unfilled
  theme(strip.text.x = element_text(size = 12, colour=black, family=serif, 
angle=00)) +
  theme(axis.text.x = element_text(size = 12, colour=black, family=serif, 
angle=90)) +
  theme(axis.text.y = element_text(size = 12, colour=black, family=serif, 
angle=00)) +
  geom_hline(yintercept=0, linetype=3) #draws dotted line at 0
  
p

###VERSION 2 

p - ggplot(mydata, aes(x = week, y = var1, fill=organism)) + #here the levels 
in organism are considered
  geom_boxplot() +
  facet_wrap(~ station, ncol = 3) +
  theme_bw() +
  theme(strip.background = element_blank())+
  ylab(var1)+
  xlab(week) +
  geom_line(stat=summary, fun.y = median,
mapping=aes(linetype=organism, group=organism))+ #must add jitter 
if using this
  geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) + 
#must be unfilled
  theme(strip.text.x = element_text(size = 12, colour=black, family=serif, 
angle=00)) +
  theme(axis.text.x = element_text(size = 12, colour=black, family=serif, 
angle=90)) +
  theme(axis.text.y = element_text(size = 12, colour=black, family=serif, 
angle=00)) +
  geom_hline(yintercept=0, linetype=3) #draws dotted line at 0
  
p
#




Anna Zakrisson Braeunlich
PhD student

Department of Ecology, Environment and Plant Sciences
Stockholm University
Svante Arrheniusv. 21A
SE-106 91 Stockholm
Sweden/Sverige

Lives in Berlin.
For paper mail:
Katzbachstr. 21
D-10965, Berlin - Kreuzberg
Germany/Deutschland

E-mail: anna.zakris...@su.se
Tel work: +49-(0)3091541281
Mobile: +49-(0)15777374888
LinkedIn: http://se.linkedin.com/pub/anna-zakrisson-braeunlich/33/5a2/51b

º`•. . • `•. .• `•. . º`•. . • `•. .• `•. .º`•. . • `•. .• 
`•. .º


From: Ista Zahn [istaz...@gmail.com]
Sent: 05 September 2013 16:02
To: Anna Zakrisson Braeunlich
Cc: r-help@r-project.org
Subject: Re: [R] ggplot2: connecting medians of boxes using facet_wrap. 
Changing median marker.

Hi Anna,

On Thu, Sep 5, 2013 at 6:13 AM, Anna Zakrisson Braeunlich
anna.zakris...@su.se wrote:
 # Dear all,

 # Thank you for taking your time.

 # What I would like to do:

 # (Run the code below to see what I am referring to)

 # I want lines connecting the medians of of the boxes. I do not want a 
 function, just a simple,
 # straight connection between points. I also would like the lines to be 
 different: lty=c(1:3)

geom_line(stat=summary, fun.y = median,
mapping=aes(linetype=organism, group=organism))


 # Furthermore, I wish to change the line/dot marking the medians to be 
 pch=c(1:3). It seems not to be so simple when using facet_wrap? I have 
 obviously missed something obvious.

geom_point(stat=summary, fun.y = median, mapping=aes(shape=organism)) +


 # Finally, I would like the boxes to be filled white and to have the legend 
 reflecting this.

Then don't tell ggplot to color them. Change

ggplot(mydata, aes(x = week, y = var1, fill = organism))

to

[R] setStatusBar function gives error message in R 3.01 under Windows 7

2013-09-05 Thread Gwen D. LaSelva


I am running R 3.01 under 64-bit Windows 7.  When I try to set the status
bar, I get an error message.  For example:

text-hello
 setStatusBar(text)
 Error in .Call(setStatusBar, text) :
   first argument must be a string (of length 1) or native symbol
reference

The related function, setWindowTitle(), appears to work just fine.

Is this a bug?  Or am I doing something wrong?  It does seem to work OK in
R 2.13.0.

Thanks,

Gwen Babcock
New York State Department of Health
Albany, NY 12237 USA
[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Problem with converting F to FALSE

2013-09-05 Thread Venkata Kirankumar

Thank you everyone
 I got actual values in my view




On Thu, Sep 5, 2013 at 7:31 PM, Keith Jewell
keith.jew...@campdenbri.co.ukwrote:

 Depending what you're doing with the data, you might want
 colClasses=c(factor,**numeric)


 On 05/09/2013 13:58, Joshua Wiley wrote:

 Hi,

 You can either manually specify colClasses or the asis argument.  See
 ?read.csv for more details.

 If you just had those two columns, something like:

   read.table(header = TRUE, text = 
   sex group
   F 1
   T 2
   , colClasses = c(character, integer))

 Cheers,

 Josh


 read.csv(file.csv, colClasses = c(character, integer))




 On Thu, Sep 5, 2013 at 5:44 AM, Venkata Kirankumar
 kiran4u2...@gmail.com  wrote:

 Hi,
 I have a peculier problem in R-Project that is when my CSV file have one
 column with all values as 'F' the R-Project converting this 'F' to FALSE.
 Can some one please suggest how to stop this convertion. Because I want
 to
 use 'F' in my calculations and show it in screen. for example my data is
 like

 sex  group
 F   1
 F   2
 F   3

 but when I use read.csv and load the csv file data is converting it to

 sex  group
 FALSE   1
 FALSE   2
 FALSE   3
 but i want it as source data like

 sex group
 F  1
 F  2
 F  3


 Thanks in advance,
 D V Kiran Kumar


 __**
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/**listinfo/r-helphttps://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/**
 posting-guide.html http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.


[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Y-axis labels as decimal numbers

2013-09-05 Thread jim holtman

So what is wrong with the y-axis?  When I run your script, things seem
right.  Can you explain what it is that you want.
Jim Holtman
Data Munger Guru

What is the problem that you are trying to solve?
Tell me what you want to do, not how you want to do it.


On Thu, Sep 5, 2013 at 9:22 AM,  mohan.radhakrish...@polarisft.com wrote:
 Hi,
  I am able to create a graph with this code but the decimal
 numbers are not plotted accurately because the ylim values are not set
 properly. x-axis is proper.

 How do I accurately set the ylim for duration.1 column ?

 Thanks,
 Mohan

 set1$duration- as.POSIXct(paste('2013-08-24', set1$duration))
 plot(set1$duration,set1$duration.1,type=b,col = blue,  ylab=, xaxt =
 'n', xlab=,las=2,lwd=2.5, lty=1,cex.axis=2.5)
 # now plot you times
 axis(1, at = set1$duration, labels = set1$duration, las = 2,cex.axis=2.5)

duration duration.1
 2  16:03:41   0.05
 3  17:03:41   0.27
 4  18:03:43   1.22
 5  19:03:45   1.51
 6  20:03:47   1.27
 7  21:03:48   1.15
 8  22:03:50   1.22
 9  23:03:52   1.27
 10 00:03:54   1.27
 11 01:03:55   1.22
 12 02:03:57   1.26
 13 03:03:59   1.57
 14 04:04:01   1.31
 15 05:04:03   1.24


 This e-Mail may contain proprietary and confidential information and is sent 
 for the intended recipient(s) only.  If by an addressing or transmission 
 error this mail has been misdirected to you, you are requested to delete this 
 mail immediately. You are also hereby notified that any use, any form of 
 reproduction, dissemination, copying, disclosure, modification, distribution 
 and/or publication of this e-mail message, contents or its attachment other 
 than by its intended recipient/s is strictly prohibited.

 Visit us at http://www.polarisFT.com

 [[alternative HTML version deleted]]

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] optim evils

Slight correction:




On Thu, Sep 5, 2013 at 7:48 AM, Bert Gunter bgun...@gene.com wrote:

 Michael:

 Your parameter specification is probably over-determined, so that you have
 an infinite set of parameter **values** that give essentially the same
 solution within numerical error. I would venture to guess that this will
 not be fixable with alternative optimizers. It is up to you to provide a
 sensible problem specification; failure to do so cannot be blamed on the
 optimizer.

 Cheers,
 Bert


 On Thu, Sep 5, 2013 at 3:23 AM, Michael Meyer spyqqq...@yahoo.com wrote:

 Thanks for all replies.
 The problem occurred in the following context:

 A Gaussian one dimensional mixture (number of constituents, locations,
 variances all unknown)
 is to be fitted to data (as starting value to or in lieu of mixtools). A
 likelihood maximization is performed.

 I'll try to destill the code so that reproducible failure of L-BFGS-B
 occurs
 and post it here.



 Michael Meyer
 [[alternative HTML version deleted]]

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide
 http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.




 --

 Bert Gunter
 Genentech Nonclinical Biostatistics

 Internal Contact Info:
 Phone: 467-7374
 Website:

 http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm





-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] binary symmetric matrix combination

Also, some of the steps could be reduced by:

names1-unique(c(colnames(m1),colnames(m2),colnames(m3),colnames(m4)))
Out3-matrix(0,length(names1),length(names1),dimnames=list(names1,names1))
lst1-sapply(paste0(m,1:4),function(x) {x1- get(x); 
x2-paste0(colnames(x1)[col(x1)],rownames(x1)[row(x1)]); match(x2,vecOut)})
lst2- list(m1,m2,m3,m4)
N- length(lst1)

 fn1- function(N,Out){
 i=1
 while(i=N){
 Out[lst1[[i]]]-lst2[[i]]
 i-i+1
 }
Out
 }
fn1(N,Out3)
#    y1 g24 c1 c2 l17 h4 s2 s30 e5 l15
#y1   0   1  1  1   1  1  1   1  1   1
#g24  1   0  0  0   0  0  0   0  0   0
#c1   1   0  0  1   1  0  0   0  0   0
#c2   1   0  1  0   1  0  0   0  0   0
#l17  1   0  1  1   0  0  0   0  0   0
#h4   1   0  0  0   0  0  1   1  0   0
#s2   1   0  0  0   0  1  0   1  0   0
#s30  1   0  0  0   0  1  1   0  0   0
#e5   1   0  0  0   0  0  0   0  0   1
#l15  1   0  0  0   0  0  0   0  1   0


 identical(Out2,fn1(N,Out3))
#[1] TRUE

A.K.


- Original Message -
From: arun smartpink...@yahoo.com
To: R help r-help@r-project.org
Cc: 
Sent: Thursday, September 5, 2013 4:09 PM
Subject: Re: binary symmetric matrix combination

Hi,
May be this helps:

m1- as.matrix(read.table(text=
y1 g24
y1 0 1
g24 1 0
,sep=,header=TRUE))

m2-as.matrix(read.table(text=y1 c1 c2 l17
 y1 0 1 1 1
 c1 1 0 1 1
 c2 1 1 0 1
 l17 1 1 1 0,sep=,header=TRUE))
m3- as.matrix(read.table(text=y1 h4    s2 s30
 y1 0 1 1 1
 h4 1 0 1 1
 s2 1 1 0 1
 s30 1 1 1 0,sep=,header=TRUE))
m4- as.matrix(read.table(text=y1 e5 l15
 y1 0 1 1
e5 1 0 1
l15 1 1 0,sep=,header=TRUE))

###desired output: at some place the label is s2 and at other s29.  I used 
s2 for consistency
Out1- as.matrix(read.table(text=y1 g24 c1 c2 l17 h4 s2 s30 e5 l15
y1 0 1 1 1 1 1 1 1 1 1
g24 1 0 0 0 0 0 0 0 0 0
c1 1 0 0 1 1 0 0 0 0 0
c2 1 0 1 0 1 0 0 0 0 0
l17 1 0 1 1 0 0 0 0 0 0
h4 1 0 0 0 0 0 1 1 0 0
s2 1 0 0 0 0 1 0 1 0 0
s30 1 0 0 0 0 1 1 0 0 0
e5 1 0 0 0 0 0 0 0 0 1
l15 1 0 0 0 0 0 0 0 1 0,sep=,header=TRUE))


names1-unique(c(colnames(m1),colnames(m2),colnames(m3),colnames(m4)))
Out2-matrix(0,length(names1),length(names1),dimnames=list(names1,names1))
vec1- paste0(colnames(m1)[col(m1)],rownames(m1)[row(m1)])
vecOut- paste0(colnames(Out2)[col(Out2)],rownames(Out2)[row(Out2)])
Out2[match(vec1,vecOut)]- m1
vec2- paste0(colnames(m2)[col(m2)],rownames(m2)[row(m2)])
Out2[match(vec2,vecOut)]- m2
vec3- paste0(colnames(m3)[col(m3)],rownames(m3)[row(m3)])
Out2[match(vec3,vecOut)]- m3
vec4- paste0(colnames(m4)[col(m4)],rownames(m4)[row(m4)])
Out2[match(vec4,vecOut)]- m4
 all.equal(Out1,Out2)
#[1] TRUE
 Out2
    y1 g24 c1 c2 l17 h4 s2 s30 e5 l15
y1   0   1  1  1   1  1  1   1  1   1
g24  1   0  0  0   0  0  0   0  0   0
c1   1   0  0  1   1  0  0   0  0   0
c2   1   0  1  0   1  0  0   0  0   0
l17  1   0  1  1   0  0  0   0  0   0
h4   1   0  0  0   0  0  1   1  0   0
s2   1   0  0  0   0  1  0   1  0   0
s30  1   0  0  0   0  1  1   0  0   0
e5   1   0  0  0   0  0  0   0  0   1
l15  1   0  0  0   0  0  0   0  1   0


A.K.



I have the following binary labeled matrices with different dimensions 
(2x2, 3x3, 4x4) which I need to create in R as seen below: 

        y1     g24 
y1     0    1 
g2 4    1    0 
                                
         y1    c1    c2    l17 
 y1    0    1    1    1 
 c1    1    0    1    1 
 c2    1    1    0    1 
 l17    1    1    1    0 
                                
       y1    h4    s2     s30 
 y1    0    1    1    1 
 h4    1    0    1    1 
 s29    1    1    0    1 
 s30    1    1    1    0 
                                
        y1     e5    l15 
 y1    0    1    1 
e5    1    0    1 
l15    1    1    0 

Then, I need to combine them to achieve the following result: 

        y1    g24    c1    c2    l17    h4    s29    s30    e5    l15 
y1    0    1    1    1    1    1    1    1    1    1 
g24    1    0    0    0    0    0    0    0    0    0 
c1    1    0    0    1    1    0    0    0    0    0 
c2    1    0    1    0    1    0    0    0    0    0 
l17    1    0    1    1    0    0    0    0    0    0 
h4    1    0    0    0    0    0    1    1    0    0 
s29    1    0    0    0    0    1    0    1    0    0 
s30    1    0    0    0    0    1    1    0    0    0 
e5    1    0    0    0    0    0    0    0    0    1 
l15    1    0    0    0    0    0    0    0    1    0 

Your help would be very much appreciated. 

ps. if the matrices don't appear correctly, please notice that all values 
different from 0 and 1 are row and column names 

Thank You!

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

[R] xyplot() with discontinuous x-axis variable

2013-09-05 Thread Rich Shepard


  My xyplot() with superposed multiple condiions looks better with lines
than with points (it's easier to see changes over time with the lines). But,
there are gaps in the years (the x axis) for which there are data to be
plotted. For example, there are data for years 2004-2006 and 2010-2012, but
not for 2007-2009. I would like to have the lines for only the two groups
with data.

  Reading ?xyplot suggests that the group attribute might do the job but I
do not see how to write the equation.

  Is it possible to plot with lines on discontinuous data or should I use
large, solid circles for each year's data points?

Rich

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Poly Correlations

2013-09-05 Thread John Fox

Dear Michael,

Please see comments below, interspersed with your questions:

On Wed, 4 Sep 2013 22:18:57 -0400
 Michael Hacker mhac...@nycap.rr.com wrote:
 Dear Colleagues,
 
  
 
 I'm working on a Delphi study comparing perceptions of high school
 technology teachers and university engineering educators about the
 importance of concepts about engineering for HS students to learn as part of
 their fundamental education. I'm actually doing this as part of my Ph.D.
 
 The survey items (n=37) are categorized into five scales: design, human
 values, modeling, resources, and systems thinking. I'm seeking to determine
 the reliability of these scales and of the overall survey instrument. Since
 I'm working with ordinal data, Chronbach's Alpha probably isn't the best
 statistical tool to use.
 
  
 
 I've literally spent several days learning my way around R-project but am
 struggling with procedures and interpretations. 
 
  
 
 I'm aware that there is now a plug-in for R for SPSS that can be downloaded
 ( http://www-01.ibm.com/support/docview.wss?uid=swg21477550
 http://www-01.ibm.com/support/docview.wss?uid=swg21477550 and
 http://gruener.userpage.fu-berlin.de/Essentials%20for%20R%20Installation%20
 Instructions_21.pdf
 http://gruener.userpage.fu-berlin.de/Essentials%20for%20R%20Installation%20I
 nstructions_21.pdf). Just learned that today and I downloaded
 PolyCorrelations.zip from
 https://www.ibm.com/developerworks/community/files/app?lang=en#/file/9f47f9a
 0-7793-4ad5-8bb7-d3fd1a028e44 
 
  
 
 I've gotten as far as loading Rcmdr and running some analyses - (Statistics,
 dimensional analysis, scale reliability) and I've generated this output:
 
  
 
 Reliability deleting each item in turn:
 
 Alpha   Std.Alpha   r(item, total)
 
 design  0.84450.8490 0.7629
 
 humanvalues   0.85260.8541 0.7170
 
 modeling  0.85110.8546 0.7271
 
 resources0.87120.8757 0.6328
 
 systems   0.84610.8498 0.7488
 
  
 
 I now would sincerely appreciate some help. At the age of 70, never having
 studied programming, the meaning of these statistics is not apparent.
 
 For example, I'm not clear if either of these three statistics are Ordinal
 Alpha. Since I'm working with Likert scale items, my advisor suggested that
 I seek an alternative to Chronbach's Alpha to determine reliability.

The table seems self-explanatory to me: it includes Chronbach's alpha and alpha 
for standardized items with each item deleted in turn, along with the 
correlation of each item with the total of the other items. All of this is 
described if you press the Help button in the Reliability dialog for the Rcmdr.

The computation isn't really appropriate for ordinal items (unless you plan to 
treat the ordinal items as numeric).

 
  
 
 So far, here are the steps I have taken:
 
 I've searched the FAQs
 
 Searched specifically for answers on the Web
 
 Played with the software for hours
 
 Read the accompanying documentation.
 
 Downloaded and installed Rcmdr
 
 Downloaded and installed PolyCorrelations. 
 
  
 
 I tried running PolyCorrelations  but I get a message that states that this
 requires the Polychor and Gclus libraries. I tried to install them into the
 R console, but no luck. 

As far as I know, this is no Polychor package on CRAN, though there is a 
polycor package, which will compute polychoric and polyserial correlations. 
These could be used to calculate reliability for ordinal items, I suppose, 
though not, to my knowledge, with the Rcmdr.

 
  
 
 I'd also be pleased to work with someone-on-one on a consulting basis if
 someone has the time and inclination.  Hoping to find an individual who
 knows SPSS and R.

It's unclear to me what SPSS has to do with all this.

Best,
 John


John Fox
McMaster University
Hamilton, Ontario, Canada
http://socserv.mcmaster.ca/jfox/


 
  
 
 Thanks very sincerely for considering this request.
 
  
 
 Michael
 
  
 
  
 
  
 
 END OF MESSAGE  
 
  
 
 Michael Hacker, Co-Director
 
 Hofstra University Center for STEM Education Research
 
 Ph: 518-724-6437
 
 Cell: 518-229-7300
 
 Fax: 518-434-6783
 
 URL: www.Hofstra.edu/CSR
 
  
 
 
   [[alternative HTML version deleted]]
 
 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] ANN: Course Data Mining with R in Spain

2013-09-05 Thread ernesto villarino

We do it tomorrow if you have some time.good night!
On Sep 3, 2013 5:04 PM, Soledad De Esteban Trivigno [via R]
ml-node+s789695n4675267...@n4.nabble.com wrote:

Dear colleague:

Registration is open for the course CLASSIFICATION AND REGRESSION TREES
AND
NEURAL NETWORKS WITH R - Second Edition.

INSTRUCTORS: Dr. LlorenÃ§ Badiella (UAB, Spain), Dr. Joan Valls
(Biomedical
Research Institute of Lleida, Spain) and Dr. Montserrat MartÃnez-Alonso
(Biomedical Research Institute of Lleida, Spain).

DATES: November 4-7, 2013; 24 teaching hours.

PLACE: Premises of Sabadell of the Institut CatalÃ de Paleontologia
Miquel
Crusafont, Sabadell, Barcelona (Spain).

Organized by: Transmitting Science and the Institut CatalÃ¡ de
Paleontologia
Miquel Crusafont.

More information: http://www.transmittingscience.org/cart_with_r.htm or
writing
to [hidden email] http://user/SendEmail.jtp?type=nodenode=4675267i=0

The main goal of the methods such as CART (Classification and Regression
Trees),
is to model and predict one response variable explained by a set of
dependent
variables. This methods can be particularly effective to model
interactions
between explanatory variables. On the other hand, as a statistical model,
a
neural network is based on linear and non-linear combinations of
explanatory
variables that interact with other combinations to predict or explain an
outcome
variable. Both CART and neural networks methods can provide good results
to
explain or predict an outcome variable, particularly when the number of
interactions is important. Nevertheless, these techniques also tend to
over-fit
the data and a validation of the models is required. ROC methods,
including a
sensitivity/specificity analyses and/or external validations can be
performed to
assess the consistency of these techniques. Applications cover a wide
range of
problems, including species classification in biology, prediction of the
prognosis of a patient in biomedicine, etc.

Please feel free to distribute this information between your colleagues if
you
consider it appropriate.

With best regards

Soledad De Esteban-Trivigno, PhD.
Academic Director
[hidden email] http://user/SendEmail.jtp?type=nodenode=4675267i=1
Transmitting Science
www.transmittingscience.org

__
[hidden email] http://user/SendEmail.jtp?type=nodenode=4675267i=2mailing
list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

--
If you reply to this email, your message will be added to the discussion
below:

http://r.789695.n4.nabble.com/ANN-Course-Data-Mining-with-R-in-Spain-tp4675267.html
To start a new topic under R help, email
ml-node+s789695n789696...@n4.nabble.com
To unsubscribe from R, click
herehttp://r.789695.n4.nabble.com/template/NamlServlet.jtp?macro=unsubscribe_by_codenode=789695code=dmlsbGFyaW5vLmVybmVzdG9AZ21haWwuY29tfDc4OTY5NXwxOTIwMjI1Njcz
.
NAMLhttp://r.789695.n4.nabble.com/template/NamlServlet.jtp?macro=macro_viewerid=instant_html%21nabble%3Aemail.namlbase=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespacebreadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml

--
View this message in context:
http://r.789695.n4.nabble.com/ANN-Course-Data-Mining-with-R-in-Spain-tp4675267p4675489.html
Sent from the R help mailing list archive at Nabble.com.
[[alternative HTML version deleted]]

[R] Matrix randomization

2013-09-05 Thread levan mumladze

Dear friends

I am very new to R and no experience in programming at all.
I need to generate random swaps of binary matrix in a way that row and
column totals remain constant. and for each derived matrix calculate site
weighted richness.
Could you suggest most appropriate function (package) to do this? Or could
you suggest a text book where such things will be explained?

Thank you in advance
Levan

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] sparse PCA using nsprcomp package

2013-09-05 Thread Christian Sigg

Hi John

I am currently traveling and have sporadic net access, I therefore can only 
answer briefly. It's also quite late, I hope what follows still makes sense...

 For regular PCA by prcomp(), we can easily calculate the percent of total 
 variance explained by the first k PCs by using cumsum(obj$sdev^2) because 
 these PCs are independent of each other so you can simply add up the variance 
 of these PCs. For sparse PCA, as far as I understand, the generated PCs are 
 not independent of each other anymore, so you can not simply add up variances 
 to calculate percentage of variance explained by the first k PCs. For 
 example, in the package of elasticnet where spca() also performs sparse PCA, 
 one of the output from spca() is pev for percent explained variation which 
 is based on so-called adjusted variance that adjusted for the fact that 
 these variances of PCs are not independent anymore.

You are correct that measuring explained variance is more involved with sparse 
(or non-negative) PCA, because the principal axes no longer correspond to 
eigenvectors of the covariance matrix, and are usually not even orthogonal.

The next update for the 'nsprcomp' package is almost done, and one of the 
changes will concern the reported standard deviations. In the current version 
(0.3), the standard deviations are computed from the scores matrix X*W, where X 
is the data matrix and W is the (pseudo-)rotation matrix consisting of the 
sparse loadings. Computing variance this way has the advantage that 'sdev' is 
consistent with the scores matrix, but it has the disadvantage that some of the 
explained variance is counted more than once because of the non-orthogonality 
of the principal axes. One of the symptoms of this counting is that the 
variance of a later component can actually exceed the variance of an earlier 
component, which is not possible in regular PCA.

In the new version of the package, 'sdev' will report the _additional_ standard 
deviation of each component, i.e. the variance not explained by the previous 
components. Given a basis of the space spanned by the previous PAs, the 
variance of the PC is computed after projecting the current PA to the 
ortho-complement space of the basis. This procedure reverts back to standard 
PCA if no sparsity or non-negativity constraints are enforced on the PAs.

 My question is for nsprcomp, how can I calculate percent explained variation 
 by using sdev when I know these PCs are not independent of each other?

The new version of the package will do it for you. Until then, you can use 
something like the following function

asdev - function(X, W) {
nc - ncol(W)
sdev - numeric(nc)
Q - qr.Q(qr(W))
Xp - X
for (cc in seq_len(nc)) {
sdev[cc] - sd(Xp%*%W[ , cc])
Xp - Xp - Xp%*%Q[ , cc]%*%t(Q[ , cc])   
}
return(sdev)
}

to compute the additional variances for given X and W.

The package documentation will explain the above in some more detail, and I 
will also have a small blog post which compares the 'nsprcomp' and 'spca' 
routine from the 'elasticnet' package on the 'marty' data from the EMA package.

Best regards
Christian

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Assessing temporal correlation in GAM with irregular time steps

2013-09-05 Thread Gavin Simpson

On 3 September 2013 16:10, Worthington, Thomas A
thomas.worthing...@okstate.edu wrote:
 Dear Gavin

 Thank you for the very detailed response. I had started to go down the route 
 of fitting a correlation structure via gamm.

 I tried applying your code to my data but returned the error
 Error in corCAR1(~ID | SiteCode1971) : parameter in CAR(1) structure must be 
 between 0 and 1

Sorry, that is my fault, I keep forgetting that you need to specify
the formula argument, the first argument of corCAR1() is the value of
the correlation parameter if you want to specify it. So try:

corCAR1(form = ~ID | SiteCode1971)

I do this (get that error) all the time myself.
 I set the 'bar' in your code to the sample ID (basically a number between 1 
 and 192) but I wasn't sure if this was what you meant in relation to 
 'ordering of the samples'

That is not that useful as you need to give the software something
about when the samples occur in time, otherwise it doesn't have the
information needed to properly model the decay in correlation with
time.

You need to give it the observation time, however you measured it.

HTH

G

 Best wishes
 Tom

 -Original Message-
 From: Gavin Simpson [mailto:ucfa...@gmail.com]
 Sent: Tuesday, September 03, 2013 3:17 PM
 To: Worthington, Thomas A
 Cc: r-help@r-project.org
 Subject: Re: [R] Assessing temporal correlation in GAM with irregular time 
 steps

 It is possible, but you can't use the discrete time or classical stochastic 
 trend models (or evaluate using the ACF). Also, why do you care to do this 
 with regard to DoY? The assumption of the model relates to the residuals, so 
 you should check those for residual autocorrelation.

 As you are using `mgcv::gam` you could also use `mgcv::gamm` which can then 
 leverage the correlation structures from the nlme package, which has spatial 
 correlation structures (and you can think of time as a 1-d spatial 
 direction). The package also has a `corCAR1()` correlation structure which is 
 the continuous-time analogue of the AR(1). Fitting via `gamm()` will also 
 allow you to use the `Variogram()` function from the nlme package to assess 
 the model residuals for residual autocorrelation.

 For example you could compare the two fits

 m0 - gamm(Length ~ s(DOY, by = SiteCode) + SiteCode, data = foo, method = 
 REML)
 m1 - gamm(Length ~ s(DOY, by = SiteCode) + SiteCode, data = foo, method = 
 REML,
 correlation = corCAR1( ~ bar | SiteCode))

 where `foo` is the object that contains the variables mentioned in the call, 
 and `bar` is the variable (in `foo)` that indicates the ordering of the 
 samples. Notice that I nest the CAR(1) within the two respective Sites, but 
 do note IIRC that this fits the same residual correlation structure to both 
 sites' residuals (i.e. there is 1 CAR(1) process, not two separate ones).

 require(nlme)
 anova(m0$lme, m1$lme)

 will perform a likelihood ratio test on the two models.

 If you have residual autocorrelation, do note that the smooth for DoY may be 
 chosen to be more complex than is appropriate (it might be fitting the 
 autocorrleated noise), so you may want to fix the degrees of freedom for the 
 smoother at some a priori chosen value and use this same value when fitting 
 both m0 and m1, or at the very least set an upper limit on the complexity of 
 the DoY smooth, say via s(DoY, by = SiteCode, k = 5).

 Finally, as a length = 0 insect makes no sense, the assumption of Gaussian 
 (Normal) errors may be in trouble with your data; apart from their strictly 
 positive nature, the mean-variance relationship of the data may not follow 
 that of the assumptions for the errors. You can move to a GLM (GAM) to 
 account for this but things get very tricky with the correlation structures 
 (you can use gamm() still but fitting then goes via glmmPQL() in the MASS 
 package a thence to lme()).

 If you just want to fit a variogram to something, there are a large number of 
 spatial packages available for R, several of which can fit variograms to 
 data, though you will need to study their respective help files for how to 
 use them. As for the input data, often the time/date of sampling encoded as a 
 numeric will be sufficient input, but you will need to check individual 
 functions for what they require.
 I would check out the Spatial Task View on CRAN.

 HTH

 G

 On 28 August 2013 14:26, Worthington, Thomas A 
 thomas.worthing...@okstate.edu wrote:
 I have constructed a GAM using the package mgcv to test whether the
 lengths of an emerging insect (Length) varies with day of the year
 (DOY) and between two sites (SiteCode). The data are collected at
 irregular time steps ranging from 2 days to 20 days between samples.
 The GAM takes the form

 M3 - gam(Length ~s(DOY, by = SiteCode) + SiteCode)

 As the data are a time series I would like to test for temporal 
 autocorrelation. I have read that it is not possible to use the 
 autocorrelation function (ACF) due to the

[R] get syntax for inherited environments

2013-09-05 Thread ivo welch

quick question.  how do I search up the calling environments until I find a
variable?

 x=function() { m=22; y() }
 y=function() { z() }
 z=function() { mget(m, inherits=TRUE, ifnotfound=m not found) }
 x()
$m
[1] m not found

from the perspective of z(), function x is not an enclosing environment.
 do I write a while loop to look back, or is there a standard R function
that searches all calling environments until it finds one that works?

regards,

/iaw

Ivo Welch (ivo.we...@gmail.com)

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] sparse PCA using nsprcomp package

2013-09-05 Thread array chip

HI Christian,

Thanks so much for the detailed explanation! I look forward to the new release 
of nsprcomp package! At the meantime, I will use the function below for 
calculation of adjusted standard deviation. I have 2 more questions, hope you 
can shed some lights on:

1). Assume now I can calculate these adjusted standard deviation from sparse 
PCA, should the percent variation explained by each sparse PC be calculated 
using the sum of all these adjusted variance (i.e. square of the adjusted 
standard deviation) as the denominator (then these percent variation explained 
will always add up to 1 if all sparse PCs are counted, or using the sum of the 
PC variances estimated by REGULAR PCA as the denominator (then, adding up all 
PCs may not be equal to 1)?

2). How do you choose the 2 important parameters in nsprcomp(), ncomp and k? If 
for example, my regular PCA showed that I need 20 PCs to account for 80% of the 
variation in my dataset, does it mean I should set ncomp=20? And then what 
about any rules setting the value of k?

3). Would you recommend nscumcomp() or nsprcomp() in general?

Thanks so much again,

John



 From: Christian Sigg r-h...@sigg-iten.ch

Cc: r-help@r-project.org 
Sent: Thursday, September 5, 2013 2:43 PM
Subject: Re: [R] sparse PCA using nsprcomp package


Hi John

I am currently traveling and have sporadic net access, I therefore can only 
answer briefly. It's also quite late, I hope what follows still makes sense...

 For regular PCA by prcomp(), we can easily calculate the percent of total 
 variance explained by the first k PCs by using cumsum(obj$sdev^2) because 
 these PCs are independent of each other so you can simply add up the variance 
 of these PCs. For sparse PCA, as far as I understand, the generated PCs are 
 not independent of each other anymore, so you can not simply add up variances 
 to calculate percentage of variance explained by the first k PCs. For 
 example, in the package of elasticnet where spca() also performs sparse PCA, 
 one of the output from spca() is pev for percent explained variation which 
 is based on so-called adjusted variance that adjusted for the fact that 
 these variances of PCs are not independent anymore.

You are correct that measuring explained variance is more involved with sparse 
(or non-negative) PCA, because the principal axes no longer correspond to 
eigenvectors of the covariance matrix, and are usually not even orthogonal.

The next update for the 'nsprcomp' package is almost done, and one of the 
changes will concern the reported standard deviations. In the current version 
(0.3), the standard deviations are computed from the scores matrix X*W, where X 
is the data matrix and W is the (pseudo-)rotation matrix consisting of the 
sparse loadings. Computing variance this way has the advantage that 'sdev' is 
consistent with the scores matrix, but it has the disadvantage that some of the 
explained variance is counted more than once because of the non-orthogonality 
of the principal axes. One of the symptoms of this counting is that the 
variance of a later component can actually exceed the variance of an earlier 
component, which is not possible in regular PCA.

In the new version of the package, 'sdev' will report the _additional_ standard 
deviation of each component, i.e. the variance not explained by the previous 
components. Given a basis of the space spanned by the previous PAs, the 
variance of the PC is computed after projecting the current PA to the 
ortho-complement space of the basis. This procedure reverts back to standard 
PCA if no sparsity or non-negativity constraints are enforced on the PAs.

 My question is for nsprcomp, how can I calculate percent explained variation 
 by using sdev when I know these PCs are not independent of each other?

The new version of the package will do it for you. Until then, you can use 
something like the following function

asdev - function(X, W) {
    nc - ncol(W)
    sdev - numeric(nc)
    Q - qr.Q(qr(W))
    Xp - X
    for (cc in seq_len(nc)) {
        sdev[cc] - sd(Xp%*%W[ , cc])
        Xp - Xp - Xp%*%Q[ , cc]%*%t(Q[ , cc])  
    }
    return(sdev)
}

to compute the additional variances for given X and W.

The package documentation will explain the above in some more detail, and I 
will also have a small blog post which compares the 'nsprcomp' and 'spca' 
routine from the 'elasticnet' package on the 'marty' data from the EMA package.

Best regards
Christian
[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] binary symmetric matrix combination

HI,
No problem.
I think you didn't run the `vecOut` after adding the new matrix.  `lst1` is 
based on `vecOut`
For example:
m5- as.matrix(read.table(text=y1 e6 l16
 y1 0 1 1
e6 1 0 1
l16 1 1 0,sep=,header=TRUE))
names1-unique(c(colnames(m1),colnames(m2),colnames(m3),colnames(m4), 
colnames(m5)))
Out3-matrix(0,length(names1),length(names1),dimnames=list(names1,names1))
lst1-sapply(paste0(m,1:5),function(x) {x1- get(x); 
x2-paste0(colnames(x1)[col(x1)],rownames(x1)[row(x1)]); match(x2,vecOut)})
 lst1
#$m1
#[1]  1  2 11 12
#
#$m2
 #[1]  1  3  4  5 21 23 24 25 31 33 34 35 41 43 44 45
#
#$m3
 #[1]  1  6  7  8 51 56 57 58 61 66 67 68 71 76 77 78
#
#$m4
#[1]   1   9  10  81  89  90  91  99 100
#
#$m5
#[1]  1 NA NA NA NA NA NA NA NA  
###Here vecOut was based on Out2


lst2- list(m1,m2,m3,m4,m5)
N- length(lst1)

 fn1- function(N,Out){
 i=1
 while(i=N){
 Out[lst1[[i]]]-lst2[[i]]
 i-i+1
 }
Out
 }
fn1(N,Out3) 
#Error in Out[lst1[[i]]] - lst2[[i]] : 
#  NAs are not allowed in subscripted assignments

###Running vecOut using Out3

vecOut-paste0(colnames(Out3)[col(Out3)],rownames(Out3)[row(Out3)])  
lst1-sapply(paste0(m,1:5),function(x) {x1- get(x); 
x2-paste0(colnames(x1)[col(x1)],rownames(x1)[row(x1)]); match(x2,vecOut)})
fn1(N,Out3) 
#    y1 g24 c1 c2 l17 h4 s2 s30 e5 l15 e6 l16
#y1   0   1  1  1   1  1  1   1  1   1  1   1
#g24  1   0  0  0   0  0  0   0  0   0  0   0
#c1   1   0  0  1   1  0  0   0  0   0  0   0
#c2   1   0  1  0   1  0  0   0  0   0  0   0
#l17  1   0  1  1   0  0  0   0  0   0  0   0
#h4   1   0  0  0   0  0  1   1  0   0  0   0
#s2   1   0  0  0   0  1  0   1  0   0  0   0
#s30  1   0  0  0   0  1  1   0  0   0  0   0
#e5   1   0  0  0   0  0  0   0  0   1  0   0
#l15  1   0  0  0   0  0  0   0  1   0  0   0
#e6   1   0  0  0   0  0  0   0  0   0  0   1
#l16  1   0  0  0   0  0  0   0  0   0  1   0

A.K.

 


Thanks a lot, all the codes worked perfectly. I have an additional question on 
the last steps you mentioned. 

I wanted to add another matrix to the ones I gave as an example,
 inputing m5 worked well, however when I type the code (added colnames 
(m5), changed 1:4 with 1:5 and added m5 to list2 I get the following 
error: 

Error in Out[lst1[[i]]] - lst2[[i]] : 
  NAs are not allowed in subscripted assignments 

Below is the code: (am I doing something wrong? very many thanks again for 
helping!! 

names1-unique(c(colnames(m1),colnames(m2),colnames(m3),colnames(m4), 
colnames(m5))) 
Out3-matrix(0,length(names1),length(names1),dimnames=list(names1,names1)) 
lst1-sapply(paste0(m,1:5),function(x) {x1- get(x); 
x2-paste0(colnames(x1)[col(x1)],rownames(x1)[row(x1)]); 
match(x2,vecOut)}) 
lst2- list(m1,m2,m3,m4,m5) 
N- length(lst1) 

 fn1- function(N,Out){ 
 i=1 
 while(i=N){ 
 Out[lst1[[i]]]-lst2[[i]] 
 i-i+1 
 } 
Out 
 } 
fn1(N,Out3) 



- Original Message -
From: arun smartpink...@yahoo.com
To: R help r-help@r-project.org
Cc: 
Sent: Thursday, September 5, 2013 4:30 PM
Subject: Re: binary symmetric matrix combination

Also, some of the steps could be reduced by:

names1-unique(c(colnames(m1),colnames(m2),colnames(m3),colnames(m4)))
Out3-matrix(0,length(names1),length(names1),dimnames=list(names1,names1))
lst1-sapply(paste0(m,1:4),function(x) {x1- get(x); 
x2-paste0(colnames(x1)[col(x1)],rownames(x1)[row(x1)]); match(x2,vecOut)})
lst2- list(m1,m2,m3,m4)
N- length(lst1)

 fn1- function(N,Out){
 i=1
 while(i=N){
 Out[lst1[[i]]]-lst2[[i]]
 i-i+1
 }
Out
 }
fn1(N,Out3)
#    y1 g24 c1 c2 l17 h4 s2 s30 e5 l15
#y1   0   1  1  1   1  1  1   1  1   1
#g24  1   0  0  0   0  0  0   0  0   0
#c1   1   0  0  1   1  0  0   0  0   0
#c2   1   0  1  0   1  0  0   0  0   0
#l17  1   0  1  1   0  0  0   0  0   0
#h4   1   0  0  0   0  0  1   1  0   0
#s2   1   0  0  0   0  1  0   1  0   0
#s30  1   0  0  0   0  1  1   0  0   0
#e5   1   0  0  0   0  0  0   0  0   1
#l15  1   0  0  0   0  0  0   0  1   0


 identical(Out2,fn1(N,Out3))
#[1] TRUE

A.K.


- Original Message -
From: arun smartpink...@yahoo.com
To: R help r-help@r-project.org
Cc: 
Sent: Thursday, September 5, 2013 4:09 PM
Subject: Re: binary symmetric matrix combination

Hi,
May be this helps:

m1- as.matrix(read.table(text=
y1 g24
y1 0 1
g24 1 0
,sep=,header=TRUE))

m2-as.matrix(read.table(text=y1 c1 c2 l17
 y1 0 1 1 1
 c1 1 0 1 1
 c2 1 1 0 1
 l17 1 1 1 0,sep=,header=TRUE))
m3- as.matrix(read.table(text=y1 h4    s2 s30
 y1 0 1 1 1
 h4 1 0 1 1
 s2 1 1 0 1
 s30 1 1 1 0,sep=,header=TRUE))
m4- as.matrix(read.table(text=y1 e5 l15
 y1 0 1 1
e5 1 0 1
l15 1 1 0,sep=,header=TRUE))

###desired output: at some place the label is s2 and at other s29.  I used 
s2 for consistency
Out1- as.matrix(read.table(text=y1 g24 c1 c2 l17 h4 s2 s30 e5 l15
y1 0 1 1 1 1 1 1 1 1 1
g24 1 0 0 0 0 0 0 0 0 0
c1 1 0 0 1 1 0 0 0 0 0
c2 1 0 1 0 1 0 0 0 0 0
l17 1 0 1 1 0 0 0 0 0 0
h4 1 0 0 0 0 0 1 1 0 0
s2 1 0 0 0 0 1 0 1 0 0
s30 1 0 0 0 0 1 1 0 0 0
e5 1 0 0 0 0 0 0 0 0 1
l15 1 0 0 0 0 0 0 0 1 0,sep=,header=TRUE))

Re: [R] get syntax for inherited environments

?sys.parent  (and friends)

-- Bert


On Thu, Sep 5, 2013 at 4:37 PM, ivo welch ivo.we...@anderson.ucla.eduwrote:

 quick question.  how do I search up the calling environments until I find a
 variable?

  x=function() { m=22; y() }
  y=function() { z() }
  z=function() { mget(m, inherits=TRUE, ifnotfound=m not found) }
  x()
 $m
 [1] m not found

 from the perspective of z(), function x is not an enclosing environment.
  do I write a while loop to look back, or is there a standard R function
 that searches all calling environments until it finds one that works?

 regards,

 /iaw
 
 Ivo Welch (ivo.we...@gmail.com)

 [[alternative HTML version deleted]]

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide
 http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.




-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm

[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Y-axis labels as decimal numbers

2013-09-05 Thread Jim Lemon


On 09/05/2013 11:22 PM, mohan.radhakrish...@polarisft.com wrote:

Hi,
  I am able to create a graph with this code but the decimal
numbers are not plotted accurately because the ylim values are not set
properly. x-axis is proper.

How do I accurately set the ylim for duration.1 column ?


Hi Mohan,
I think you may have your axes mixed up. Try this and see:

set1-read.table(text=duration duration.1
16:03:41   0.05
17:03:41   0.27
18:03:43   1.22
19:03:45   1.51
20:03:47   1.27
21:03:48   1.15
22:03:50   1.22
23:03:52   1.27
00:03:54   1.27
01:03:55   1.22
02:03:57   1.26
03:03:59   1.57
04:04:01   1.31
05:04:03   1.24,header=TRUE)
set1$duration- as.POSIXct(paste(c(rep('2013-08-24',8),rep('2013-08-25',6)),
 set1$duration))
par(mar=c(10,4,4,2))
plot(set1$duration,set1$duration.1,type=b,col=blue,ylab=,xaxt='n',
 xlab=,las=2,lwd=2.5,lty=1,cex.axis=2.5)
axis(1, at = set1$duration, labels = set1$duration, las = 2,cex.axis=1)
par(mar=c(5,4,4,2))

Jim

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Looping an lapply linear regression function

2013-09-05 Thread Flavio Barros

Hello Arun. Can you provide some data? To help you better i will need a
complete reproducible example ok?


On Thu, Sep 5, 2013 at 1:49 PM, arun smartpink...@yahoo.com wrote:

 HI,
 May be this helps:
  set.seed(28)
  dat1-
 setNames(as.data.frame(matrix(sample(1:40,10*5,replace=TRUE),ncol=5)),letters[1:5])
 indx-as.data.frame(combn(names(dat1),2),stringsAsFactors=FALSE)
 res-t(sapply(indx,function(x)
 {x1-cbind(dat1[x[1]],dat1[x[2]]);summary(lm(x1[,1]~x1[,2]))$coef[,4]}))
  rownames(res)-apply(indx,2,paste,collapse=_)
  colnames(res)[2]- Coef1
  head(res,3)
 #(Intercept) Coef1
 #a_b  0.39862676 0.8365606
 #a_c  0.02427885 0.6094141
 #a_d  0.37521423 0.7578723


 #permutation
 indx2-expand.grid(names(dat1),names(dat1),stringsAsFactors=FALSE)
 #or
 indx2- expand.grid(rep(list(names(dat1)),2),stringsAsFactors=FALSE)
 indx2New- indx2[indx2[,1]!=indx2[,2],]
 res2-t(sapply(seq_len(nrow(indx2New)),function(i) {x1- indx2New[i,];
 x2-cbind(dat1[x1[,1]],dat1[x1[,2]]);summary(lm(x2[,1]~x2[,2]))$coef[,4]}))
 row.names(res2)-apply(indx2New,1,paste,collapse=_)
  colnames(res2)- colnames(res)


 A.K.


 Hi everyone,

 First off just like to say thanks to everyone´s contributions.
 Up until now, I´ve never had to post as I´ve always found the answers
 from trawling through the database. I´ve finally managed to stump
 myself, and although for someone out there, I´m sure the answer to my
 problem is fairly simple, I, however have spent the whole day infront of
  my computer struggling. I know I´ll probably get an absolute ribbing
 for making a basic mistake, or not understanding something fully, but
 I´m blind to the mistake now after looking so long at it.

 What I´m looking to do, is formulate a matrix ([28,28]) of
 p-values produced from running linear regressions of 28 variables
 against themselves (eg a~b, a~c, a~d.b~a, b~c etc...), if that makes
  sense. I´ve managed to get this to work if I just input each variable
 by hand, but this isn´t going to help when I have to make 20 matrices.

 My script is as follows;


 for (j in [1:28])
 {
  ##This section works perfectly, if I don´t try to loop it, I know
 this wont work at the moment, because I haven´t designated what j is,
 but I´m showing to highlight what I´m attempting to do.


models - lapply(varlist, function(x) {
 lm(substitute(ANS ~ i, list(i = as.name(x))), data = con.i)
   })

   abc- lapply(models, function(f) summary(f)$coefficients[,4])

   abc- do.call(rbind, abc)



 }

 I get the following error when I try to loop it...

 Error in model.frame.default(formula = substitute(j ~ i, list(i = 
 as.name(x))),
  :
   variable lengths differ (found for 'ANS') ##ÄNS being my first variable

 All variables are of the same length, with 21 recordings for each


 If anyone can suggest a method of looping, or another means
 or producing ´models´ for each of my 28 variables, without having to do
 it by hand that would be fantastic.

 Thanks in advance!!

 __
 R-help@r-project.org mailing list
 https://stat.ethz.ch/mailman/listinfo/r-help
 PLEASE do read the posting guide
 http://www.R-project.org/posting-guide.html
 and provide commented, minimal, self-contained, reproducible code.


[[alternative HTML version deleted]]

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] setStatusBar function gives error message in R 3.01 under Windows 7

2013-09-05 Thread Duncan Murdoch


On 05/09/2013 10:20 AM, Gwen D. LaSelva wrote:

I am running R 3.01 under 64-bit Windows 7.  When I try to set the status
bar, I get an error message.  For example:

text-hello
 setStatusBar(text)
 Error in .Call(setStatusBar, text) :
   first argument must be a string (of length 1) or native symbol
reference

The related function, setWindowTitle(), appears to work just fine.

Is this a bug?  Or am I doing something wrong?  It does seem to work OK in
R 2.13.0.


Could you please try the R-patched nightly build (from 
http://cran.r-project.org/bin/windows/base/rpatched.html)?  I think this 
has already been fixed.


Duncan Murdoch

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] setStatusBar function gives error message in R 3.01 under Windows 7

2013-09-05 Thread Duncan Murdoch


On 05/09/2013 10:20 AM, Gwen D. LaSelva wrote:

I am running R 3.01 under 64-bit Windows 7.  When I try to set the status
bar, I get an error message.  For example:

text-hello
 setStatusBar(text)
 Error in .Call(setStatusBar, text) :
   first argument must be a string (of length 1) or native symbol
reference

The related function, setWindowTitle(), appears to work just fine.

Is this a bug?  Or am I doing something wrong?  It does seem to work OK in
R 2.13.0.


Thanks, looks like a bug, and looks easy to fix.  Should make it into 
3.0.2 later this month.


Duncan Murdoch

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

[R] applying a univariate function for each response in a multivariate linear model (mlm)

2013-09-05 Thread Michael Friendly

After fitting a multivariate linear model (mlm), I'd like to be able to 
run or apply
a standard univariate stats:::*.lm function to each of the response 
variables,
within a function -- i.e., by operating on the mlm object, rather than 
re-running

the univariate models separately manually.

An example: extracting cooks.distance components (via 
stats:::cooks.distance.lm)


grain - c(40, 17, 9, 15, 6, 12, 5, 9)# y1
straw - c(53, 19, 10, 29, 13, 27, 19, 30)# y2
fertilizer - c(24, 11, 5, 12, 7, 14, 11, 18) # x

Fertilizer - data.frame(grain, straw, fertilizer)
# fit the mlm
mod - lm(cbind(grain, straw) ~ fertilizer, data=Fertilizer)

# run univariate regressionsand get cooks.distance
 (cookd.grain - cooks.distance(lm(grain ~ fertilizer, data=Fertilizer)))
 1  2  3  4 5  6  7
3.4436e+00 4.0957e-02 2.2733e-01 4.8605e-03 1.4073e-05 2.0479e-02 
6.4192e-02

 8
4.8383e-01
 (cookd.straw - cooks.distance(lm(straw ~ fertilizer, data=Fertilizer)))
1 2 3 4 5 6 7 8
2.0003953 0.0283225 0.0675803 0.1591198 0.0013352 0.0024076 0.0283225 
0.4672299


This is the result I want:

 data.frame(cookd.grain, cookd.straw)
  cookd.grain cookd.straw
1  3.4436e+00   2.0003953
2  4.0957e-02   0.0283225
3  2.2733e-01   0.0675803
4  4.8605e-03   0.1591198
5  1.4073e-05   0.0013352
6  2.0479e-02   0.0024076
7  6.4192e-02   0.0283225
8  4.8383e-01   0.4672299

Note that if I call cooks.distance.lm directly on the mlm object, there 
is no complaint

or warning, but the result is silently WRONG:

 # try calling cooks.distance.lm directly:  silently WRONG
  stats:::cooks.distance.lm(mod)
   grain  straw
1 3.4436e+00 0.51729792
2 1.5838e-01 0.02832250
3 2.2733e-01 0.01747613
4 1.8796e-02 0.15911979
5 1.4073e-05 0.00034527
6 7.9192e-02 0.00240762
7 6.4192e-02 0.00732414
8 1.8710e+00 0.46722985


I realize that I can also use update() on the mlm object to re-fit the 
univariate models,
but I don't know how to extract the response names from it to do this in 
a function


 coef(mod)  # multivariate
  grain   straw
(Intercept) -3.7524 -2.2965
fertilizer   1.4022  2.1409

 coef(update(mod, grain ~ .))
(Intercept)  fertilizer
-3.7524  1.4022
 coef(update(mod, straw ~ .))
(Intercept)  fertilizer
-2.2965  2.1409



--
Michael Friendly Email: friendly AT yorku DOT ca
Professor, Psychology Dept.  Chair, Quantitative Methods
York University  Voice: 416 736-2100 x66249 Fax: 416 736-5814
4700 Keele StreetWeb:   http://www.datavis.ca
Toronto, ONT  M3J 1P3 CANADA

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] Looping an lapply linear regression function

HI,
May be this helps:
 set.seed(28)
 dat1- 
setNames(as.data.frame(matrix(sample(1:40,10*5,replace=TRUE),ncol=5)),letters[1:5])
indx-as.data.frame(combn(names(dat1),2),stringsAsFactors=FALSE)
res-t(sapply(indx,function(x) 
{x1-cbind(dat1[x[1]],dat1[x[2]]);summary(lm(x1[,1]~x1[,2]))$coef[,4]}))
 rownames(res)-apply(indx,2,paste,collapse=_)
 colnames(res)[2]- Coef1
 head(res,3)
#    (Intercept) Coef1
#a_b  0.39862676 0.8365606
#a_c  0.02427885 0.6094141
#a_d  0.37521423 0.7578723


#permutation
indx2-expand.grid(names(dat1),names(dat1),stringsAsFactors=FALSE)
#or
indx2- expand.grid(rep(list(names(dat1)),2),stringsAsFactors=FALSE)
indx2New- indx2[indx2[,1]!=indx2[,2],]
res2-t(sapply(seq_len(nrow(indx2New)),function(i) {x1- indx2New[i,]; 
x2-cbind(dat1[x1[,1]],dat1[x1[,2]]);summary(lm(x2[,1]~x2[,2]))$coef[,4]}))
row.names(res2)-apply(indx2New,1,paste,collapse=_)
 colnames(res2)- colnames(res)


A.K.


Hi everyone, 

First off just like to say thanks to everyone´s contributions. 
Up until now, I´ve never had to post as I´ve always found the answers 
from trawling through the database. I´ve finally managed to stump 
myself, and although for someone out there, I´m sure the answer to my 
problem is fairly simple, I, however have spent the whole day infront of
 my computer struggling. I know I´ll probably get an absolute ribbing 
for making a basic mistake, or not understanding something fully, but 
I´m blind to the mistake now after looking so long at it. 

What I´m looking to do, is formulate a matrix ([28,28]) of 
p-values produced from running linear regressions of 28 variables 
against themselves (eg a~b, a~c, a~d.b~a, b~c etc...), if that makes
 sense. I´ve managed to get this to work if I just input each variable 
by hand, but this isn´t going to help when I have to make 20 matrices. 

My script is as follows; 


for (j in [1:28]) 
{ 
 ##This section works perfectly, if I don´t try to loop it, I know 
this wont work at the moment, because I haven´t designated what j is, 
but I´m showing to highlight what I´m attempting to do.   
  

   models - lapply(varlist, function(x) { 
    lm(substitute(ANS ~ i, list(i = as.name(x))), data = con.i) 
  }) 
  
          abc- lapply(models, function(f) summary(f)$coefficients[,4]) 
  
          abc- do.call(rbind, abc) 
  
          
  
} 

I get the following error when I try to loop it... 

Error in model.frame.default(formula = substitute(j ~ i, list(i = as.name(x))), 
 : 
  variable lengths differ (found for 'ANS') ##ÄNS being my first variable 

All variables are of the same length, with 21 recordings for each 


If anyone can suggest a method of looping, or another means 
or producing ´models´ for each of my 28 variables, without having to do 
it by hand that would be fantastic. 

Thanks in advance!!

__
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Re: [R] optim evils