Skip to content

Calculate median from counts and values

3 messages · David Finlayson, Gabor Grothendieck

Original
David Finlayson
# 
I am tangled with a syntax question. I want to calculate basic statistics  
for a large dataset provided in weights and values and I can't figure out  
an elegant way to expand the data.

For example here are the counts:
n4 n3 n2 n1 p0 p1 p2 p3  p4
1   0  0  0  1  1  3 16 55  24
2   0  0  0  0  2  8 28 47  15
3   1 17 17 13  4  5 12 24   8
...

and the values:
n4 n3 n2 n1 p0  p1   p2    p3     p4
[1,] 16  8  4  2  1 0.5 0.25 0.125 0.0625

What I want for each row is something like this (shown for row 1):

c( rep(16, 0), rep(8, 0), rep(4, 0), rep(2, 1), rep(1, 1), rep(0.5, 3),  
rep(0.25, 16), rep(0.125, 55), rep(0.0625, 24))

I am sure that this is a one-liner for an R-master, but I can't figure it  
out without a set of nested for loops iterating over each row in counts.

David
Gabor Grothendieck
#
On 5/3/05, David Finlayson <david.p.finlayson at gmail.com> wrote:
Is there supposed to be one row of values that apply to all
rows of counts or is there to be different rows of values for
different rows of counts?  Also in your example row 3 has
a different total than 1 or 2.  Is that right?

At any rate, I will assume that there is only one row of 
values and many rows of counts and that its not necessarily
true that counts sum to the same number in each row.
Then noting that  c(rep(4,1), rep(5,2), rep(6,3)) is the same
as rep(4:6, 1:3) is the same as, we have:

lapply(as.data.frame(t(counts)), rep, x = unlist(values))
David Finlayson
# 
Thanks Gabor and Phil. That did it.

I've used R for years for plotting and run-of-the-mill data analysis
(the only kind I do). But the syntax of this language has just never
clicked for me. I can't seem to advance beyond the "mostly harmless"
stage. Python is roting my brain I guess.

Again, thanks for the tips

David
On 5/3/05, Gabor Grothendieck <ggrothendieck at gmail.com> wrote: