Why is my matrix not numeric?

Question

I have a 32X32 matrix with the columns and rows representing each NFL team. Running across the rows indicates on average how much that team beat their opponent (indicated by the column name).

I created 32 vectors for each team:

nfl_data <- read.csv(file="2012NFLScoresMatrixNoNames.csv", h=T)
colnames(nfl_data) <- NULL

ari_s <- nfl_data[1,]
atl_s <- nfl_data[2,]
.
.
.
was_s <- nfl_data[32,]

a <- rbind(c(ari_s),c(atl_s),c(bal_s),c(buf_s),c(car_s),c(chi_s),c(cin_s),
    c(cle_s),c(dal_s),c(den_s),c(det_s),c(grn_s),c(hou_s),c(ind_s),
    c(jac_s),c(kc_s),c(mia_s),c(min_s),c(ne_s),c(no_s),c(nyj_s),
    c(nyg_s),c(oak_s),c(phi_s),c(pit_s),c(sd_s),c(sea_s),c(sf_s),
    c(stl_s),c(tb_s),c(tn_s),c(was_s))

a_trans <- t(a)
x <- solve(a_trans)

This is where i get an error:

Error in solve.default(a_trans) : 'a' must be a numeric matrix

> str(nfl_data)
'data.frame':   32 obs. of  32 variables:
$ : num  0 4 0 3 0 15 0 0 0 0 ...
$ : int  -4 0 0 0 4 0 0 0 -6 -6 ...
$ : num  0 0 0 0 0 0 -12.5 -8.5 -2 17 ...
$ : num  -3 0 0 0 0 0 0 -10 0 0 ...
$ : int  0 -4 0 0 0 1 0 0 5 22 ...
$ : num  -15 0 0 0 -1 0 0 0 -16 0 ...
$ : num  0 0 12.5 0 0 0 0 1.5 1 8 ...
$ : num  0 0 8.5 10 0 0 -1.5 0 3 22 ...
$ : num  0 6 2 0 -5 16 -1 -3 0 0 ...
$ : num  0 6 -17 0 -22 0 -8 -22 0 0 ...
$ : num  28 13 0 0 0 4 0 0 0 0 ...
$ : num  -14 0 0 0 0 -10.5 0 0 0 0 ...
$ : int  0 0 -30 -12 0 -7 0 0 0 -6 ...
$ : int  0 0 0 -7 0 20 0 -4 0 0 ...
$ : num  0 0 0 16 0 38 17 0 0 0 ...
$ : num  0 16 3 18 -6 0 22 23 0 21.5 ...
$ : num  3 0 0 -4.5 0 0 -4 0 0 0 ...
$ : num  -7 0 0 0 0 5 0 0 0 0 ...
$ : num  2 0 1 -15 0 0 0 0 0 -10 ...
$ : int  0 3 0 0 7 0 0 0 -3 10 ...
$ : num  -1 0 0 -0.5 0 0 0 0 0 0 ...
$ : num  0 34 19 0 -29 0 18 -14 1 0 ...
$ : num  0 3 35 0 11 0 24 3 0 22 ...
$ : num  21 13 -1 0 8 0 21 -1 10 0 ...
$ : int  0 0 0 0 0 0 -2 -4 3 12 ...
$ : num  0 24 0 0 26 0 7 1 0 9 ...
$ : num  -27 0 0 -33 -4 -6 0 0 -20 0 ...
$ : num  -17.5 0 0 -42 0 -25 0 0 0 0 ...
$ : num  -14 0 0 -3 0 17 0 0 0 0 ...
$ : int  0 -2 0 0 -6 0 0 0 6 8 ...
$ : num  0 0 0 -1 0 31 0 0 0 0 ...
$ : num  0 7 -3 0 8 0 7 -17 -8.5 0 ...

> str(a)
List of 1024
$ : num 0
$ : num 4
$ : num 0
$ : num 3
$ : num 0
$ : num 15
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num -28
$ : num 14
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num -3
$ : num 7
$ : num -2
$ : num 0
$ : num 1
$ : num 0
$ : num 0
$ : num -21
$ : num 0
$ : num 0
$ : num 27
$ : num 17.5
$ : num 14
$ : num 0
$ : num 0
$ : num 0
$ : int -4
$ : int 0
$ : int 0
$ : int 0
$ : int 4
$ : int 0
$ : int 0
$ : int 0
$ : int -6
$ : int -6
$ : int -13
$ : int 0
$ : int 0
$ : int 0
$ : int 0
$ : int -16
$ : int 0
$ : int 0
$ : int 0
$ : int -3
$ : int 0
$ : int -34
$ : int -3
$ : int -13
$ : int 0
$ : int -24
$ : int 0
$ : int 0
$ : int 0
$ : int 2
$ : int 0
$ : int -7
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num -12.5
$ : num -8.5
$ : num -2
$ : num 17
$ : num 0
$ : num 0
$ : num 30
$ : num 0
$ : num 0
$ : num -3
$ : num 0
$ : num 0
$ : num -1
$ : num 0
$ : num 0
$ : num -19
$ : num -35
$ : num 1
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num 0
$ : num 3
$ : num -3
$ : num 0
$ : num 0
  [list output truncated]
  - attr(*, "dim")= int [1:2] 32 32

I looked over and over to see if I had a comma or maybe an o instead of a 0, but I couldn't find anything. Any suggestions on how to solve this?

Answer 1

The object a is a list, not a matrix or data frame. Have a look at your code:

a <- rbind(c(ari_s),c(atl_s),c(bal_s),c(buf_s),c(car_s),c(chi_s),c(cin_s),
    c(cle_s),c(dal_s),c(den_s),c(det_s),c(grn_s),c(hou_s),c(ind_s),
    c(jac_s),c(kc_s),c(mia_s),c(min_s),c(ne_s),c(no_s),c(nyj_s),
    c(nyg_s),c(oak_s),c(phi_s),c(pit_s),c(sd_s),c(sea_s),c(sf_s),
    c(stl_s),c(tb_s),c(tn_s),c(was_s))

Here, you used c for each object. If you omit the c function calls, a will be a data frame, not a common list.

A small example:

> x <- data.frame(a=1:2, b=0.4)
> x1 <- x[1,]
> x2 <- x[2,]

> str(rbind(c(x1), c(x2)))
List of 4
 $ : int 1
 $ : int 2
 $ : num 0.4
 $ : num 0.4
 - attr(*, "dim")= int [1:2] 2 2
 - attr(*, "dimnames")=List of 2
  ..$ : NULL
  ..$ : chr [1:2] "a" "b"

> str(rbind(x1, x2))
'data.frame':   2 obs. of  2 variables:
 $ a: int  1 2
 $ b: num  0.4 0.4

The function c transforms the object of class data.frame into an object of class list .

Maybe it's much easier to use

solve(t(nfl_data))

The data frame nfl_data will be converted into a matrix due to t .

Answer 2

The informative bit when you do str(a) is this:

> str(a)
List of 1024
....

In other words, a is a list, not a matrix. The "numeric" qualifier in the error message 'a' must be a numeric matrix is there to stop people complaining when a user tries to invert the other types of matrix (logical and character). In your case, the error message means: " You didn't give me a matrix of any sort ".

The easiest way to solve this is just to transpose the data frame:

solve(t(nfl_data))

where the use of t() will convert the nfl_data data frame to a matrix via the as.matrix() coercion function. To be more explicit and if you data might include factors, then you may wish to convert yourself via data.matrix() , eg

solve(t(data.matrix(nfl_data)))

But either way, there is no need to strip out by hand each row-vector from the data frame, bind these row-wise and then transpose.