How should I interprent the singular value decomposition(SVD)?

33 Views Asked by At

Let's say that I have this data matrix, which describe measurements of temperature over 10 different weather stations:

$$H = \begin{bmatrix} -5 & -5 & -6& -10& -20& -3& -20& -25& -30& 6\\ -4 & -6 & -6& -9& -10& -2& -15& -20& -25& 6\\ -3 & -3 & -6& -10& -11& 1& -10& -15& -22& 4\\ -4 & -3 & -4& -4& -12& 2& -5& -10& -20& 7\\ -3 & -4 & -2& -1& -10& 5& 0& -5& -18& 9\\ 1 & -1& 1& 4& -4& 6& 2& 1& -15& 10\\ 0 & 0& 0& 5& -4& 7& 5& 5& -14& 15\\ 2 & 1& 1& 6& -2& 5& 10& 7& -13& 20\\ 3 & 2& 2& 7& -1& 4& 15& 10& -12& 30\\ 5 & 1& 3& 8& 0& 3& 20& 15& -5& 31 \end{bmatrix}$$

Where each column is measurement each day and each row is the weather station.

CODE:

H = [
-5  -5   -6  -10 -20  -3  -20  -25  -30 6; 
-4  -6   -6  -9  -10  -2  -15  -20  -25 6; 
-3  -3   -6  -10  -11  1  -10  -15  -22 4; 
-4  -3   -4  -4  -12  2  -5  -10  -20 7; 
-3  -4   -2  -1  -10  5  0  -5  -18 9;
1   -1  1  4  -4  6 2  1 -15 10; 
0   0  0  5  -4  7  5 5  -14 15; 
2   1  1  6  -2  5  10  7  -13 20;
3   2  2  7  -1  4  15  10  -12 30; 
5   1  3  8  0  3  20  15  -5 31]

[U, E, V] = svd(H)

But what would $U$, $E$ and $V$ mean? Well, I know that $E$ is just scalars. But I don't know what kind if scalars they are. Are they scalars of temperature? Scalars of vectors in some weird spaces?

Can some one explain what $U$, $E$ and $V$ mean in this case?

1

There are 1 best solutions below

1
On BEST ANSWER

The singular value decomposition is relevant when you see your matrix as a linear operator. If you just have rows of data, it is hard to imagine that the singular value decomposition will tell you anything about your data.