Find index where value exceeds threshold

Dear all,

I want to be able to find indices where values exceed a certain threshold. So if my variable of interest is called 'returns' and is formatted in a t x n matrix, I would like to find all indices in 'returns' that exceed a value X.

I'm working on a project that uses in- and out-of-sample stock returns to check the expediency of my model. Therefore, I don't want to just delete return outliers, but check manually if they make sense.

Thanks in advance.

Answer 1

[Andrew Reibold]

This should answer your problem! :-)

idx = find(returns > X)

Now if you have an MxN matrix, the idx returned is the single index number as if you concatenated all the rows.

Do you need help turning it into two index numbers? I can do that too.

[Image Analyst]

You mean "concatenated all the columns" instead of "concatenated all the rows". Try a simple example to see.

[Andrew Reibold]

Yes, I mean concatenated the column, you are correct

[Yoni Sher]

Isn't 'find' inefficient? is there some other function that is better?

[Image Analyst]

No. Use tic and toc to time it and you'll see it's plenty fast enough for you.

Answer 2

[Image Analyst]

Look at this simple example:

m=magic(4)  % Sample data
linearIndexes = find(m>8)  % Find elements with value more than 8
[rows, columns] = ind2sub(size(m), linearIndexes)

In the command window:

m =
    16     2     3    13
     5    11    10     8
     9     7     6    12
     4    14    15     1
linearIndexes =
     1
     3
     6
     8
    10
    12
    13
    15
rows =
     1
     3
     2
     4
     2
     4
     1
     3
columns =
     1
     1
     2
     2
     3
     3
     4
     4