How can I modify vector length?

I have several vectors to compare, however they all slightly differentiate by length. I know that first numbers of vectors are zeros. Can I cut those number to get vectors of same length? I don't want to pad vectors with zeros, because in this case I wouldn't be able to calculate an error between vectors?

Or may be there is a way to calculate error ratio between two vectors, even if they have different length?

[Jan]

How do you defined an "error" ratio of vectors, when they have a different length? Why does zero padding not help, but zero cropping does? And if zero cropping does, why don't you do it?

Note that it is hard to suggest a solution, which matches your needs exactly, because the readers cannot guess, what "I have several vectors" mean. Do you mean 20 vectors store in a cell array, or 20'000 vectors in different variables?

[DenisK]

I have 4 vectors stored in different variables. Error evaluation I want to perform by mean absolute percentage error (MAPE) method, and it's code looks like this:

Error1 = abs(mean(abs(clin - ccirc')./clin)*100)

However, as a result I get NaN. Even after cropping zeros.

[Jan]

Please post the data. If the inputs contain NaNs, the result must be NaN also.

[DenisK]

The data is an mp3 file, so I cannot attache it here. Vectors that I built are result of few convolution methods, that I need to compare (linear, circular, overlap save and overlap add). I took linear convolution result as a reference vector.

Answer 1

[Jan]

If cropping the zeros solves the problem, crop the zeros:

v = [0,0,1,2,3,4]
v = v(find(v, 1):length(v))