2009 April

FDIC closed 4 more banks and N…

04-24-2009 by dan

FDIC closed 4 more banks and NCUA closed a credit union… things aren’t getting better yet

Tags: Twitter | Comments (0)

once again, failed to spot Cha…

04-17-2009 by dan

once again, failed to spot Champ as i stared out over lake champlain

Tags: Twitter | Comments (0)

R: tapply Output Formatting

04-13-2009 by dan

I am often frustrated with the output of tapply. If you aren’t outputting 1 number for each group you are summarizing, the output comes back in a crazy list format that is tough to use.

Here’s an example use of tapply:

# setup data for tapply
dt = data.frame(bucket=rep(1:4,25),val=rnorm(100))
fn = function(x) {
  ret = c(unname(quantile(x,probs=seq(.25,.75,.25),na.rm=T)),mean(x,na.rm=T))
}
a = tapply(dt$val,dt$bucket,fn)

This code will give you quartiles and a mean for each of the groups (identified by the “bucket” variable).

You can see that the answers are actually there:

> a[1]
$`1`
[1] -1.1868349 -0.4585898  0.1100985 -0.4729515
> a[2]
$`2`
[1] -0.50152398 -0.06482399  0.95285490  0.09977271

But they are associated with the list items in each location. Unfortunately, this will not work:

b = as.data.frame(a)

If you could do this, you would have a nice usable data frame:

b = rbind(a[[1]],a[[2]],a[[3]],a[[4]])

But who wants to type out each element of the list everytime you want to construct such a table? Do you always know how many elements are in that list?

The solution is to use the function do.call. This function takes as arguments a function name and some data, and it constructs a call like the one above. Here is the syntax for our example:

b = do.call(rbind,a)

This will construct the call to rbind based on the fact that a is a list of items. It is equivalent to calling b = rbind(a[[1]],a[[2]],a[[3]],a[[4]]), but it takes care of all the nuances for you!

For good measure, you may want to bring along the names, too:

row.names(b) = row.names(a)
colnames(b) = colnames(a)

This will come in very handy for tapply output, but I am sure it will also come in handy for other list-based functions.

Comments (1)

reading about asthma as an all…

04-08-2009 by dan

reading about asthma as an allergy to cockroaches: http://tinyurl.com/cyy9f3

Tags: Twitter | Comments (0)

iphones and blackberry storms:…

04-07-2009 by dan

iphones and blackberry storms: neat portable devices, or just covered in greasy fingerprints?

Tags: Twitter | Comments (0)

if you haven't been watching j…

04-02-2009 by dan

if you haven’t been watching jim the realtor, you should be: http://www.bubbleinfo.com/

Tags: Twitter | Comments (0)

amazing! http://litlurl.net/14…

04-01-2009 by dan

amazing! http://litlurl.net/141a

Tags: Twitter | Comments (0)