dotchart for matrix data

Readers,

I am trying to use the function dotchart. The data is:
testdot
category values1 values2 values3 values4
1        a      10      27      56     709
2        b       4      46      47     208
3        c       5      17      18     109
4        d       6      50      49     308

The following error occurs
dotchart(testdot,groups=testdot[,2])
Error in dotchart(testdot, labels = testdot[, 1], groups = testdot[, 2]) :
        'x' must be a numeric vector or matrix

According to my understanding (clearly wrong!) of the documentation
for dotchart (accessed from the manual in section 'graphics'), columns
of data can be selected by 'groups' for subsequent plotting. The
objective is to be able to create a dot chart where each row is
labelled according to the row names in the 'category' column and two
columns can be selected, e.g. 'values1' and 'values2'. Then I tried:
testdot1<-testdot[,1]
testdot2<-testdot[,2]
testdot3<-testdot[,3]
dotchart(c(testdot2,testdot3),labels=testdot1)
A graph is produced, but not as expected. Instead of 4 rows labelled
(descending order from top row) a,b,c,d, and each row containing two
data points, the graph shows 8 rows (?) with the top 4 rows
un-labelled and the bottom 4 rows labelled (descending order) d,c,b,a
and each row shows only 1 datum point.

How do I specify the order of labelling of the rows?
How do I write correct commands to obtain a dot chart with 4 rows,
each row containing the two (or if required three) data points?

Thanks in advance.

Readers,

I am trying to use the function dotchart. The data is:

testdot
 category values1 values2 values3 values4
1        a      10      27      56     709
2        b       4      46      47     208
3        c       5      17      18     109
4        d       6      50      49     308

The following error occurs

dotchart(testdot,groups=testdot[,2])
Error in dotchart(testdot, labels = testdot[, 1], groups = testdot[,  
2]) :
       'x' must be a numeric vector or matrix

According to my understanding (clearly wrong!) of the documentation
for dotchart (accessed from the manual in section 'graphics'), columns
of data can be selected by 'groups' for subsequent plotting.
The misunderstanding is in how you see the grouping information versus  
how R expects it. R generally expects such data in what is called  
"long" format, i.e. there will be one values columns and a category  
column. There are various ways to change the arrangement of your data.  
The function stack(), the function reshape(), or probably most  
commonly the function melt from reshape2 being the typical chosen  
routes.
The
objective is to be able to create a dot chart where each row is
labelled according to the row names in the 'category' column and two
columns can be selected, e.g. 'values1' and 'values2'. Then I tried:

testdot1<-testdot[,1]
testdot2<-testdot[,2]
testdot3<-testdot[,3]
dotchart(c(testdot2,testdot3),labels=testdot1)
See if this is more to your liking:

require(reshape)   # I'm not sure why I have reshape_0.8.3 rather than  
reshape2 loaded
                    # I'm pretty sure Hadley would prefer that people  
use pkg:reshape2
  mdot <- melt(dot)
dotchart(mdot$value, groups=mdot$category, labels=mdot$variable)
# OR more readable
with(mdot, dotchart(value, groups=category, labels=variable)  )

I'm not sure I got the roles of "values" and "category" correct, but  
it should be a simple matter to switch them in the dotcghart call if  
that is your pleasuRe.
A graph is produced, but not as expected. Instead of 4 rows labelled
(descending order from top row) a,b,c,d, and each row containing two
data points, the graph shows 8 rows (?) with the top 4 rows
un-labelled and the bottom 4 rows labelled (descending order) d,c,b,a
and each row shows only 1 datum point.

How do I specify the order of labelling of the rows?
How do I write correct commands to obtain a dot chart with 4 rows,
each row containing the two (or if required three) data points?
David Winsemius, MD
West Hartford, CT
David Winsemius <dwinsemius <at> comcast.net> writes:

On Dec 18, 2010, at 7:01 AM, e-letter wrote:

Readers,

I am trying to use the function dotchart. The data is:

testdot
 category values1 values2 values3 values4
1        a      10      27      56     709
2        b       4      46      47     208
3        c       5      17      18     109
4        d       6      50      49     308

The following error occurs

dotchart(testdot,groups=testdot[,2])
Error in dotchart(testdot, labels = testdot[, 1], groups = testdot[,  
2]) :
       'x' must be a numeric vector or matrix

According to my understanding (clearly wrong!) of the documentation
for dotchart (accessed from the manual in section 'graphics'), columns
of data can be selected by 'groups' for subsequent plotting.

Following up on David's response:

d <- read.table(textConnection("category values1 values2 values3 values4
1        a      10      27      56     709
2        b       4      46      47     208
3        c       5      17      18     109
4        d       6      50      49     308"),
                header=TRUE)

## Something like this is probably as close as you can get with
## stock 'dotchart' -- it does *not* (as far as I can tell) put
## different points on the same line, just groups lines

dotchart(as.matrix(d[,-1]),labels=as.character(d[,1]))
dotchart(as.matrix(d[,c("values1","values2")]),labels=as.character(d[,1]))

## reshaping data:
library(reshape)
mdot <- melt(d)

## using the lattice package

library(lattice)
dotplot(value~category,groups=variable,data=mdot)
dotplot(value~variable,groups=category,data=mdot,auto.key=TRUE,
        scales=list(y=list(log=10)))

## you could also use ggplot2 ...
On Dec 18, 2010, at 7:01 AM, e-letter wrote:

Readers,

I am trying to use the function dotchart. The data is:

testdot
 category values1 values2 values3 values4
1        a      10      27      56     709
2        b       4      46      47     208
3        c       5      17      18     109
4        d       6      50      49     308

The following error occurs

dotchart(testdot,groups=testdot[,2])
Error in dotchart(testdot, labels = testdot[, 1], groups = testdot[,
2]) :
       'x' must be a numeric vector or matrix

According to my understanding (clearly wrong!) of the documentation
for dotchart (accessed from the manual in section 'graphics'), columns
of data can be selected by 'groups' for subsequent plotting.
The misunderstanding is in how you see the grouping information versus
how R expects it. R generally expects such data in what is called
"long" format, i.e. there will be one values columns and a category
column. There are various ways to change the arrangement of your data.
The function stack(), the function reshape(), or probably most
commonly the function melt from reshape2 being the typical chosen
routes.

Reshape and melt are not installed (version251) so for this task
manual rearrangement data is easier.
require(reshape)
Loading required package: reshape
[1] FALSE
Warning message:
there is no package called 'reshape' in: library(package, lib.loc =
lib.loc, character.only = TRUE, logical = TRUE,
library(reshape)
Error in library(reshape) : there is no package called 'reshape'
mdot<-melt(dot)
Error: could not find function "melt"

However before doing so why this is relevant because of the
alternative creation objects 'testdot1'. Aren't these objects
suitable, since a (undesireable) graph was produced?
The
objective is to be able to create a dot chart where each row is
labelled according to the row names in the 'category' column and two
columns can be selected, e.g. 'values1' and 'values2'. Then I tried:

testdot1<-testdot[,1]
testdot2<-testdot[,2]
testdot3<-testdot[,3]
dotchart(c(testdot2,testdot3),labels=testdot1)
See if this is more to your liking:

require(reshape)   # I'm not sure why I have reshape_0.8.3 rather than
reshape2 loaded
                    # I'm pretty sure Hadley would prefer that people
use pkg:reshape2
  mdot <- melt(dot)
dotchart(mdot$value, groups=mdot$category, labels=mdot$variable)
# OR more readable
with(mdot, dotchart(value, groups=category, labels=variable)  )

I'm not sure I got the roles of "values" and "category" correct, but
it should be a simple matter to switch them in the dotcghart call if
that is your pleasuRe.

I don't have dotcghart either.
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
David Winsemius <dwinsemius <at> comcast.net> writes:

On Dec 18, 2010, at 7:01 AM, e-letter wrote:

Readers,

I am trying to use the function dotchart. The data is:

testdot
 category values1 values2 values3 values4
1        a      10      27      56     709
2        b       4      46      47     208
3        c       5      17      18     109
4        d       6      50      49     308

The following error occurs

dotchart(testdot,groups=testdot[,2])
Error in dotchart(testdot, labels = testdot[, 1], groups = testdot[,
2]) :
       'x' must be a numeric vector or matrix

According to my understanding (clearly wrong!) of the documentation
for dotchart (accessed from the manual in section 'graphics'), columns
of data can be selected by 'groups' for subsequent plotting.

 Following up on David's response:

d <- read.table(textConnection("category values1 values2 values3 values4
1        a      10      27      56     709
2        b       4      46      47     208
3        c       5      17      18     109
4        d       6      50      49     308"),
               header=TRUE)

## Something like this is probably as close as you can get with
## stock 'dotchart' -- it does *not* (as far as I can tell) put
## different points on the same line, just groups lines

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
dotchart(as.matrix(d[,-1]),labels=as.character(d[,1]))
dotchart(as.matrix(d[,c("values1","values2")]),labels=as.character(d[,1]))

## reshaping data:
library(reshape)
mdot <- melt(d)

## using the lattice package

library(lattice)
dotplot(value~category,groups=variable,data=mdot)
dotplot(value~variable,groups=category,data=mdot,auto.key=TRUE,
       scales=list(y=list(log=10)))

## you could also use ggplot2 ...
?ggplot2
No documentation for 'ggplot2' in specified packages and libraries:
you could try 'help.search("ggplot2")'; seems I need to retrieve this
package first. Thanks for the suggestion.
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]
I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

Peter Ehlers

dotchart(as.matrix(d[,-1]),labels=as.character(d[,1]))
dotchart(as.matrix(d[,c("values1","values2")]),labels=as.character(d[,1]))

## reshaping data:
library(reshape)
mdot<- melt(d)

## using the lattice package

library(lattice)
dotplot(value~category,groups=variable,data=mdot)
dotplot(value~variable,groups=category,data=mdot,auto.key=TRUE,
        scales=list(y=list(log=10)))

## you could also use ggplot2 ...

?ggplot2
No documentation for 'ggplot2' in specified packages and libraries:
you could try 'help.search("ggplot2")'; seems I need to retrieve this
package first. Thanks for the suggestion.

______________________________________________
R-help at 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.
On 2010-12-18 07:50, e-letter wrote:
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

Unfortunately I have been unable to use the "melt" package; 3 mirrors
have failed to install. What I don't understand is why the objects
created are not in the correct format, otherwise how would the plot be
created?

On 18/12/2010, Peter Ehlers <ehlers at ucalgary.ca> wrote:
On 2010-12-18 07:50, e-letter wrote:
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

Unfortunately I have been unable to use the "melt" package; 3 mirrors
have failed to install. What I don't understand is why the objects
created are not in the correct format, otherwise how would the plot be
created?
There is no melt package. There is a reshape package and a reshape2  
package, either of which will have melt as a function.

______________________________________________
R-help at 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.
David Winsemius, MD
West Hartford, CT
On 2010-12-18 07:50, e-letter wrote:
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

For the benefit of other novices, this is what I did (gnu/linux):

sign-in to a command terminal as root
start R
install.packages("reshape2")

In another terminal as normal user

require(reshape2)
mdot<-melt(testdot)
dotplot(value~category,groups=variable,data=mdot)

The resultant graph shows the categories on the abscissa and there
does not seem to be a way of selecting two columns from the original
matrix. It is not like the graph cited in the hyperlink quoted above,
but at least I have successfully followed instructions!

I changed the axes with the command:

dotplot(category~value,groups=variable,data=mdot)

I believe there is a command to select only certain rows of data which
I think will achieve the desired graph, if unable I'll ask again, so
thank you all.

On 18/12/2010, Peter Ehlers <ehlers at ucalgary.ca> wrote:
On 2010-12-18 07:50, e-letter wrote:
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

For the benefit of other novices, this is what I did (gnu/linux):

sign-in to a command terminal as root
start R
install.packages("reshape2")

In another terminal as normal user

require(reshape2)
mdot<-melt(testdot)
dotplot(value~category,groups=variable,data=mdot)

The resultant graph shows the categories on the abscissa and there
does not seem to be a way of selecting two columns from the original
matrix. It is not like the graph cited in the hyperlink quoted above,
but at least I have successfully followed instructions!
If you want to see a worked example of a dotplot that has two separate  
data series on the same set of lines then look at Fig 10.18 in  
Sarkar's "Lattice". The illustration and code are all at the book's  
website. I don't think you did a very good job of explaining in text  
what you expected to see (at least on my reading of it more than once)  
but this may do what you hoped:

dotplot(as.character(mdot$category) ~ mdot$value, labels=mdot$variable)
I changed the axes with the command:

dotplot(category~value,groups=variable,data=mdot)

I believe there is a command to select only certain rows of data which
I think will achieve the desired graph, if unable I'll ask again, so
thank you all.

______________________________________________
R-help at 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.
David Winsemius, MD
West Hartford, CT
On 18/12/2010, Peter Ehlers<ehlers at ucalgary.ca>  wrote:
On 2010-12-18 07:50, e-letter wrote:
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

For the benefit of other novices, this is what I did (gnu/linux):

sign-in to a command terminal as root
start R
install.packages("reshape2")

In another terminal as normal user

require(reshape2)
mdot<-melt(testdot)
dotplot(value~category,groups=variable,data=mdot)

The resultant graph shows the categories on the abscissa and there
does not seem to be a way of selecting two columns from the original
matrix. It is not like the graph cited in the hyperlink quoted above,
but at least I have successfully followed instructions!
It seems to me that you're getting exactly the plot referred
to by the link, except that you have a factor with 4 levels, not 2.
I changed the axes with the command:

dotplot(category~value,groups=variable,data=mdot)

I believe there is a command to select only certain rows of data which
I think will achieve the desired graph, if unable I'll ask again, so
thank you all.
See if the 'subset=' argument works for you:

  dotplot(category ~ value, groups = variable, data = mdot,
          subset = {variable %in% c("values1", "values2")},
          pch = c(1,3), cex = 1.5)

Peter Ehlers
On 18/12/2010, Peter Ehlers <ehlers at ucalgary.ca> wrote:
On 2010-12-18 07:50, e-letter wrote:
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

If one set of value ranges for 10-20 and another set ranges form
1000-1500, how to adjust the graph such that:

with categories on the ordinate (y-axis), can the bottom abscissa
(x-axis bottom) be set with a scale suitable for the data set of range
1000-1500?
can the top abscissa (x-top) be set with a scale 10-20?
or is it better practice to change the scale of the data set 1000-1500
by two orders of magnitude?
how is it possible to control the order of the category variables? For
example, if the graph shows:

a
b
c
d

Is it possible to change to

a
d
b
c

Thanks
On 18/12/2010, e-letter <inpost at gmail.com> wrote:
On 18/12/2010, Peter Ehlers <ehlers at ucalgary.ca> wrote:
On 2010-12-18 07:50, e-letter wrote:
Ben Bolker
Sat, 18 Dec 2010 07:07:24 -0800
[... snip ...]

I am trying to create a chart like this
(http://www.b-eye-network.com/images/content/Fig4_3.jpg); so this is
not possible using R?
That looks an awful lot like what lattice's dotplot would
produce. So: have you tried dotplot() as Ben has suggested?

A csv file is imported:
testdot
category values
1        a     10
2        b     44
3        c     51
4        d     65
5        a     27
6        b     64
7        c     71
8        d     49

A dotplot is produced with the command:

dotplot(category~values,data=testdot)

The csv file is adjusted to move the values of 'a':
testdot
category values
1        b     44
2        c     51
3        d     65
4        a     10
5        b     64
6        c     71
7        d     49
8        a     27

Why  does the order of the categories not change from d,c,b,a to
a,d,c,b (descending from top)?
An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20101222/e30b23e5/attachment.pl>