how can i compute the stander diviation and confidence intervales for a matrix?

i have matrix 51 rows and71 columns and i would like to compute and plot the result of std also i need to compute the confidence intervals 95%?

also after computing the std can i use imagesc command in order to see the range of std for the matrix ?

thank you

Answer 1

[the cyclist]

If you have the Image Processing Toolbox, you can use the std2 function to compute the standard deviation of all the elements of a matrix. If not, you can use std from base MATLAB, but you need to make a vector out of the matrix first.

A = magic(5);

std2(A)

ans = 7.3598

std((A(:)))

ans = 7.3598

Sorry, but I didn't really understand your second question.

[RADWAN A F ZEYADI]

thanks ithe second one is how can i compute confidenc intervals for 95%

[the cyclist]

Oh, I guess I meant I didn't understand the third question (about imagesc).

You need to be more precise about the confidence interval. I can think of at least two things you could mean:

1. What is the interval of values such that 95% of your matrix values lie within that interval
2. You've calculated the mean matrix value, and you want to calculate the error of the mean, with 95th percentile confidence interval

Can you describe in a bit more detail what you want?

[RADWAN A F ZEYADI]

i mean i want to see the values of confidence interval for specifc position for example for the column number 60 because this matrix is the true model and i have another one for the predicted model with the same dimension so the aim to make a comparison for specific position to see the fitting between them by using confidence interval i hope i was clear thank you

[RADWAN A F ZEYADI]

about imagesc command i want to see the std range along my picture the colors of std

[the cyclist]

You keep trickling out small bits of information that you are trying to solve, and it is making me concerned that my small answers are not actually what you need. Now, it sounds like you are trying to make some kind of statistical goodness-of-fit calculation. My answers are not precisely what you need.

I recommend starting a new question for this. Upload your data here, and explain carefully what problem you are actually trying to solve. Don't guess at the statistics you need, because frankly it sounds like you don't understand that aspect very well, and you need some help on that.

[RADWAN A F ZEYADI]

@the cyclist

i wan to estimate confidence intervals

[the cyclist]

You can use the prctile function to calculate a confidence interval. For example, if X is a vector then

x_lower = prctile(X, 2.5);
x_upper = prctile(X,97.5);

will give the lower and upper limits of the interval that contains 95% of the values.

[RADWAN A F ZEYADI]

thank you but they are 2 vectors so can i compute for both of them sementanuesly?

[the cyclist]

N = 1000;
X1 = randn(N,1);
X2 = randn(N,1) + 100;
X = [X1, X2];
x_lower = prctile(X, 2.5)

x_lower = 1×2

-1.9542 97.9812

x_upper = prctile(X,97.5)

x_upper = 1×2

1.9782 101.7530

Here I have defined two vectors (of equal length), put them into a matrix, and calculated the confidence interval for both vectors simultaneously.