Covariance Matrix Of a Random Vector

[kadimzamanlar]

nice lectures professor, thanks for sharing with us.

[slatun]

You are GOD !!!

[tritile]

Do you know how to use this function on MatLab?

I'm trying to find the covariance of a vector, but it gives me a single number, not a matrix

[npatwari]

@tritile You need many realizations of that vector in order to estimate the covariance matrix. So, however you measured or came up with that one vector, do it N times, and then send the cov() function all of the realizations of the random vector together (in a matrix). Then Matlab will return a covariance matrix.

[tritile]

@npatwari

I'm sorry to bother you

For example, I have the vector V=[ 2; 3; 5]

so I'll make my matrix M=[2 3 5; 2 3 5; 2 3 5]

and then cov(M)?

I guess I dind't understand what you mean, because I tried this way and MatLab gave me one matrix filled by zeros

I'd be glad if you help me

thanks

[armanrainy]

@tritile

Hi(sorry about my English). I think you didn't get the meaning of random variables. when we say X it means we have several matrices and it makes our data set. for example, X is a 3*1 matrix which means three different experiment results about 3 parameters( like density,temperature,pressure). When we say X it means we have done this experiment N times (so we have n 3*1 matrices) and we want to find the covariance matrix.your matlab result is caused by having just 1 observation.

[tritile]

@armanrainy Thank you for the answer. I got what you told. Actually, my knowledge’s about statistics are not very good. But I solved the problem using another way. I realized that I didn’t need do find the covariance matrix in my case. But, anyway, thank you for the help!