Better way to get a frequency table for continuous data (R)?

Question

With df:

df <- data.frame(value=abs(rnorm(100, 25, 5)), status=sample(0:1,100,replace=T))
df$value[sample(1:100,5)] <- NA

I need to get a frequency (percentage) table (better return a matrix) like the following:

value | status(0)  status(1)
----------------------------
 <=25 |  23 (23%)   20 (20%)
  >25 |  27 (27%)   25 (25%)
   NA |   3  (3%)    2  (2%)

I can do this using:

br <- seq(0, 50, 25)
with(df, summary(cut(value[status==0], br, labels=br[-1],
     include.lowest=T, ordered_result=T)))
with(df, summary(cut(value[status==1], br, labels=br[-1],
     include.lowest=T, ordered_result=T)))

But would there be a one-time way to return a matrix as above? Thanks!

John Colby · Accepted Answer · 2011-11-02T20:09:10.910

14

df$value.cut = cut(df$value, breaks=c(0, 25, 100))

> with(df, table(value.cut, status, useNA='ifany'))
          status
value.cut   0  1
  (0,25]   26 19
  (25,100] 26 24
  <NA>      3  2

(Of course this can be combined into 1 line if you want, but I left it as 2 here for better readability.)

EDIT: And if you want a table of proportions, formatted as frequencies, you can do:

df.tab = with(df, table(value.cut, status, useNA='ifany'))
df.tab[,] = paste(df.tab, ' (', 100*prop.table(df.tab), '%)', sep='')

> df.tab
          status
value.cut  0        1       
  (0,25]   26 (26%) 19 (19%)
  (25,100] 26 (26%) 24 (24%)
  <NA>     3 (3%)   2 (2%)

edited Nov 02 '11 at 20:09

answered Nov 02 '11 at 16:33

John Colby

22,169
4
57
69

4

And, if you don't want to specify left and right truncation points, set `breaks=c(-Inf, 25, Inf)`. – Josh O'Brien Nov 02 '11 at 16:40
Nice tip. I always forget that, myself. – John Colby Nov 02 '11 at 16:43
Thanks! Should have known table() better. :) – Rock Nov 02 '11 at 16:45

score 2 · Answer 2 · answered Nov 02 '11 at 17:05

2

Another solution using reshape2.

library(reshape2)
dcast(df, cut(value, breaks = c(0, 25, 100)) ~ status)

answered Nov 02 '11 at 17:05

Ramnath

54,439
16
125
152

Better way to get a frequency table for continuous data (R)?

2 Answers2