R help - Dec 2010 - Hmisc label function applied to data frame

Hello,

I'm attempting to create a data frame with correlations between every pair
of variables in a data frame, so that I can then sort by the value of the
correlation coefficient and see which pairs of variables are most strongly
correlated.

The sm2vec function in the corpcor library works very nicely as shown here:

library(Hmisc)
library(corpcor)

# Create example data
x1 = runif(50)
x2 = runif(50)
x3 = runif(50)
d = data.frame(x1=x1,x2=x2,x3=x3)
label(d$x1) = "Variable x1"
label(d$x2) = "Variable x2"
label(d$x3) = "Variable x3"

# Get correlations
cormat = cor(d)

# Get vector form of lower triangular elements
cors = sm2vec(cormat,diag=F)
inds = sm.index(cormat,diag=F)

# Create a data frame
var1 = dimnames(cormat)[[1]][inds[,1]]
var2 = dimnames(cormat)[[2]][inds[,2]]
lbl1 = label(d[,var1])
lbl2 = label(d[,var2])
cor_df = data.frame(Var1=lbl1,Var2=lbl2,Cor=cors)

The issue that I run into is when trying to get the labels in lbl1 and
lbl2.  I get the warning:

In mapply(FUN = label, x = x, default = default, MoreArgs = list(self TRUE),  :
  longer argument not a multiple of length of shorter

My usage of label seems ambiguous since the data frame could also a label
attached to it, aside from labels attached to variables within the data
frame.  However, the code above does work, with the warning.  Aside from
using a loop to get the label of one variable at a time, is there any other
way of getting the labels for all variables in the data frame?

Also, if there is a better way to achieve my goal of getting the
correlations between all variable pairs, I'd love to know.

Thanks in advance for any responses!

--Krishna

	[[alternative HTML version deleted]]

KT_rfan wrote:
> 
> 
> I'm attempting to create a data frame with correlations between every
pair
> of variables in a data frame, so that I can then sort by the value of the
> correlation coefficient and see which pairs of variables are most strongly
> correlated.
> 
> The sm2vec function in the corpcor library works very nicely as shown
> here:
> 
> library(Hmisc)
> library(corpcor)
> 
> # Create example data
> x1 = runif(50)
> x2 = runif(50)
> x3 = runif(50)
> d = data.frame(x1=x1,x2=x2,x3=x3)
> label(d$x1) = "Variable x1"
> label(d$x2) = "Variable x2"
> label(d$x3) = "Variable x3"
> .. rest of code omitted
> 
This has nothing to do with Hmisc and corpcor. I things get confusing,
simplify and use str(). 

What you wanted to "label" the columns is "names", or,
probably better
named, "colnames". Note that your way of labeling converts the column
to a
Class "labelled", which is not what function take for breakfeast.


d = data.frame(x1=runif(10),x2=runif(10))
label(d)  # This alone gives the error message
# x1 x2 
# "" "" "
# Warning message:
#In mapply(FUN = label, x = x, default = default, MoreArgs = list(self TRUE),  :
#  longer argument not a multiple of length of shorter
str(d)
# data.frame':   10 obs. of  2 variables:
# $ x1: num  0.1353 0.7234 0.0266 0.074 0.2391 ...
# $ x2: num  0.833 0.573 0.136 0.395 0.308 ...
label(d$x1) = "Variable x1"
str(d)

#'data.frame':   10 obs. of  2 variables:
# $ x1:Class 'labelled'  atomic [1:10] 0.1353 0.7234 0.0266 0.074 0.2391
...
#  .. ..- attr(*, "label")= chr "Variable x1"
# $ x2: num  0.833 0.573 0.136 0.395 0.308 ...


# Labeling columns, the correct way
d = data.frame(x1=runif(10),x2=runif(10))
str(d)
names(d) = c("Var1","Var2")
str(d)




-- 
View this message in context:
http://r.789695.n4.nabble.com/Hmisc-label-function-applied-to-data-frame-tp3069784p3070777.html
Sent from the R help mailing list archive at Nabble.com.

Hello again,

I have found that if I use sapply, I do not get a warning, i.e.,
  lbl1 = sapply(d[,var1],label)
works correctly and gives no warning.

I'm sorry this did not occur to me earlier, my apologies!

--Krishna

On Thu, Dec 2, 2010 at 11:36 AM, Krishna Tateneni
<tateneni@gmail.com>wrote:
> Hello,
>
> I'm attempting to create a data frame with correlations between every
pair
> of variables in a data frame, so that I can then sort by the value of the
> correlation coefficient and see which pairs of variables are most strongly
> correlated.
>
> The sm2vec function in the corpcor library works very nicely as shown here:
>
> library(Hmisc)
> library(corpcor)
>
> # Create example data
> x1 = runif(50)
> x2 = runif(50)
> x3 = runif(50)
> d = data.frame(x1=x1,x2=x2,x3=x3)
> label(d$x1) = "Variable x1"
> label(d$x2) = "Variable x2"
> label(d$x3) = "Variable x3"
>
> # Get correlations
> cormat = cor(d)
>
> # Get vector form of lower triangular elements
> cors = sm2vec(cormat,diag=F)
> inds = sm.index(cormat,diag=F)
>
> # Create a data frame
> var1 = dimnames(cormat)[[1]][inds[,1]]
> var2 = dimnames(cormat)[[2]][inds[,2]]
> lbl1 = label(d[,var1])
> lbl2 = label(d[,var2])
> cor_df = data.frame(Var1=lbl1,Var2=lbl2,Cor=cors)
>
> The issue that I run into is when trying to get the labels in lbl1 and
> lbl2.  I get the warning:
>
> In mapply(FUN = label, x = x, default = default, MoreArgs = list(self >
TRUE),  :
>   longer argument not a multiple of length of shorter
>
> My usage of label seems ambiguous since the data frame could also a label
> attached to it, aside from labels attached to variables within the data
> frame.  However, the code above does work, with the warning.  Aside from
> using a loop to get the label of one variable at a time, is there any other
> way of getting the labels for all variables in the data frame?
>
> Also, if there is a better way to achieve my goal of getting the
> correlations between all variable pairs, I'd love to know.
>
> Thanks in advance for any responses!
>
> --Krishna
>
	[[alternative HTML version deleted]]