How can I extract values from a vector element that includes NaN elements?

Hi, Let's suppose that we have the following vector

V = [3 4 5 NaN NaN NaN 10 NaN 15 45 95 32 65 NaN NaN]

I would like to extract the numerical values and create new variables from those values.

V1 = [3 4 5]

V2 = [10]

V3 = [15 45 95 32 65]

I think is important to mention that the number of new variables (V1,V2 and V3) should be based on the number of "events" encounter in the vector. In this case 3 "events".

Also, the first vector value can be a number or NaN element

Regards

Answer 1

[Guillaume]

Note that creating numbered variables is an extremely bad idea (search the forum or faq to know why). We'll be creating a cell array that can easily be indexed instead.

Assuming that V does not start by NaN:

V = [3 4 5 NaN NaN NaN 10 NaN 15 45 95 32 65 NaN NaN];  %demo data
grouplength = diff([0 find(diff(isnan(V))) numel(V)]);  %find length of non-nan and nan runs
Vgrouped = mat2cell(V, 1, grouplength);   %split according to runs
Vgrouped = Vgrouped(1:2:end)   %get rid of nan runs, assumes that 1st run is not nan

Your V1, V2, V3 are Vgrouped{1}, Vgrouped{2}, Vgrouped{3} instead. Something that can be easily looped over unlike numbered variables.

[Jose Valles]

Thank you for your reply!

You are absolutely correct regarding the numbered variables.

How much it would change the code line if the V variable starts by NaN?

I am telling you this because I want to use this code line in different dataset and sometimes it can start by NaN or not.

[Guillaume]

If it starts with NaN you'd start the indexing at 2:

Vgrouped = Vgrouped(2:2:end)

To cater for both cases:

Vgrouped = Vgrouped(1+isnan(V(1)):2:end)