R - How to fill a matrix by columns -

i trying fill columns of matrix subsets of built-in data frame. resulting matrix should have dimensions 16 11 , each subset 16 integers long.

i have written following loop:

a <- unique(dnase$run) z <- matrix(data =na, ncol=11, nrow =16) (i in a) {    z[,i] <- subset(dnase$density, dnase$run==i) } 

and obtain following error:

error in [<-(*tmp*, , i, value = c(0.017, 0.018, 0.121, 0.124, 0.206, : no 'dimnames' attribute array

could kindly explain confusion comes from?

many in advance!

cheap solution

since dnase data.frame ordered run factor, can form desired output matrix simple call matrix():

matrix(dnase$density,16); ##        [,1]  [,2]  [,3]  [,4]  [,5]  [,6]  [,7]  [,8]  [,9] [,10] [,11] ##  [1,] 0.017 0.045 0.070 0.011 0.035 0.086 0.094 0.054 0.032 0.052 0.047 ##  [2,] 0.018 0.050 0.068 0.016 0.035 0.103 0.092 0.054 0.043 0.094 0.057 ##  [3,] 0.121 0.137 0.173 0.118 0.132 0.191 0.182 0.152 0.142 0.164 0.159 ##  [4,] 0.124 0.123 0.165 0.108 0.135 0.189 0.182 0.148 0.155 0.166 0.155 ##  [5,] 0.206 0.225 0.277 0.200 0.224 0.272 0.282 0.226 0.239 0.259 0.246 ##  [6,] 0.215 0.207 0.248 0.206 0.220 0.277 0.273 0.222 0.242 0.256 0.252 ##  [7,] 0.377 0.401 0.434 0.364 0.385 0.440 0.444 0.392 0.420 0.439 0.427 ##  [8,] 0.374 0.383 0.426 0.360 0.390 0.426 0.439 0.383 0.395 0.439 0.411 ##  [9,] 0.614 0.672 0.703 0.620 0.658 0.686 0.686 0.658 0.624 0.690 0.704 ## [10,] 0.609 0.681 0.689 0.640 0.647 0.676 0.668 0.644 0.705 0.701 0.684 ## [11,] 1.019 1.116 1.067 0.979 1.060 1.062 1.052 1.043 1.046 1.042 0.994 ## [12,] 1.001 1.078 1.077 0.973 1.031 1.072 1.035 1.002 1.026 1.075 0.980 ## [13,] 1.334 1.554 1.629 1.424 1.425 1.424 1.409 1.466 1.398 1.340 1.421 ## [14,] 1.364 1.526 1.479 1.399 1.409 1.459 1.392 1.381 1.405 1.406 1.385 ## [15,] 1.730 1.932 2.003 1.740 1.750 1.768 1.759 1.743 1.693 1.699 1.715 ## [16,] 1.710 1.914 1.884 1.732 1.738 1.806 1.739 1.724 1.729 1.708 1.721 

this of course depends on aforementioned ordering, can verified call rle():

do.call(data.frame,rle(levels(dnase$run)[dnase$run])); ##    lengths values ## 1       16      1 ## 2       16      2 ## 3       16      3 ## 4       16      4 ## 5       16      5 ## 6       16      6 ## 7       16      7 ## 8       16      8 ## 9       16      9 ## 10      16     10 ## 11      16     11 

robust solution

if don't want depend on ordering, can use reshape() follows, , nice bonus of column names, if want that:

reshape(cbind(dnase[c('run','density')],id=ave(c(dnase$run),dnase$run,fun=seq_along)),dir='w',timevar='run')[-1]; ##    density.1 density.2 density.3 density.4 density.5 density.6 density.7 density.8 density.9 density.10 density.11 ## 1      0.017     0.045     0.070     0.011     0.035     0.086     0.094     0.054     0.032      0.052      0.047 ## 2      0.018     0.050     0.068     0.016     0.035     0.103     0.092     0.054     0.043      0.094      0.057 ## 3      0.121     0.137     0.173     0.118     0.132     0.191     0.182     0.152     0.142      0.164      0.159 ## 4      0.124     0.123     0.165     0.108     0.135     0.189     0.182     0.148     0.155      0.166      0.155 ## 5      0.206     0.225     0.277     0.200     0.224     0.272     0.282     0.226     0.239      0.259      0.246 ## 6      0.215     0.207     0.248     0.206     0.220     0.277     0.273     0.222     0.242      0.256      0.252 ## 7      0.377     0.401     0.434     0.364     0.385     0.440     0.444     0.392     0.420      0.439      0.427 ## 8      0.374     0.383     0.426     0.360     0.390     0.426     0.439     0.383     0.395      0.439      0.411 ## 9      0.614     0.672     0.703     0.620     0.658     0.686     0.686     0.658     0.624      0.690      0.704 ## 10     0.609     0.681     0.689     0.640     0.647     0.676     0.668     0.644     0.705      0.701      0.684 ## 11     1.019     1.116     1.067     0.979     1.060     1.062     1.052     1.043     1.046      1.042      0.994 ## 12     1.001     1.078     1.077     0.973     1.031     1.072     1.035     1.002     1.026      1.075      0.980 ## 13     1.334     1.554     1.629     1.424     1.425     1.424     1.409     1.466     1.398      1.340      1.421 ## 14     1.364     1.526     1.479     1.399     1.409     1.459     1.392     1.381     1.405      1.406      1.385 ## 15     1.730     1.932     2.003     1.740     1.750     1.768     1.759     1.743     1.693      1.699      1.715 ## 16     1.710     1.914     1.884     1.732     1.738     1.806     1.739     1.724     1.729      1.708      1.721 

note technically above object data.frame, can coerce matrix as.matrix().

explanation of error

the reason why code failing follows. first, notice dnase$run vector ordered factor:

class(dnase$run); ## [1] "ordered" "factor" 

your a variable therefore ordered factor, unique values dnase$run.

now, when use for-loop iterate on factor (ordered or otherwise), uses levels (character strings) iteration value (as opposed integer enumeration values stored internally). demo:

for (i in factor(letters[1:5])) print(i); ## [1] "a" ## [1] "b" ## [1] "c" ## [1] "d" ## [1] "e" 

thus, i loop variable being assigned levels character strings of dnase$run. and, since z matrix has no dimnames, trying index columns character string failing error message "no 'dimnames' attribute array".