How to continue while loop if NaN value is present in cell array

I am using a cell 1x54 cell array containing 600x8 double values. In this array, i am looking at one column in particular for each array cell (column 5) and i want to count the number of seconds it takes until the data crossess a threshold value.

I am able to simply do this with a while loop except my data has NaN values which are causing the while loop to stop. I want to count these NaN values and keep the loop progressing until that threshold is reached.

I have a row vector 'Timetosteady' which i want to count the steps into (eventually making it a 1x54 vector) and already have calculated a threshold "MinC" which is a 1x54 row vector containing threshold values.

I am encountering two issues. Firstly using this loop trying to calculate across all cells in the cell array:

TimetosteadyC=NaN(1,54);
i=1;
for j=1:54
while ((C{1,j}(i,5))< MinC(i) || isnan(C{1,j}(i,5)))
    TimetosteadyC(j)=i+1;
    i=i+1;
end
 end

i receive the error "Index exceeds the number of array elements (54).". There defintely are 54 columns in my array (for example if i use C{1,54}(1,5) i receive a value) so why is this occuring?

Secondly, if i use this code only assessing in the first cell array:

 TimetosteadyC=NaN(1,54);
i=1;
for j=1
while ((C{1,j}(i,5))< MinC(i) || isnan(C{1,j}(i,5)))
    TimetosteadyC(j)=i+1;
    i=i+1;
end

The code does skip the first NaN value seen in the column i am looking at within the array but then stops before the threshold has been reached (the 'Timetosteady' value spits out as 25 when it should be closer to 150)

I would really appreciate guidance on how to best create a loop in a cell array that sequentially counts the values in a specific column until the threshold has been crossed while including NaN values.

btw i am on R2019b

Thanks

Answer 1

[Matt J]

Since all your arrays C{j} are the same size, it seems unnecessary to carry the data around in a cell array at all. You could just concatenate them into a 3D array:

c=cat(3,C{:});

Now, you can generate the result using simple numeric array indexing. Also, no loops are required:

c5=c(:,5,:);

[~,Timetosteady]=min( c5<reshape(MinC,1,1,[]) | isnan(c5) );

Timetosteady=Timetosteady(:);

[Luke McCauley]

Hi Matt,

Using a 3d array is a great idea, thank you for your help. I am not very familiar with using this in Matlab yet so i may need a bit more help.

When i run the code c and c5 correctly run but i have the following issue appear:

Array dimensions must match for binary array op.

Error in SteadyStateCount (line 45)
[~,Timetosteady]=min( c5<MinC(:) | isnan(c5) )

Is this because my MinC is still a 2D array do you think? Do you have suggestons on how to fix this?

Thanks

[Matt J]

Try it now with my latest edit.

Incidentally, in your original code, you have MinC(i) when it should really be MinC(j).

[Luke McCauley]

Ah yes thats worked! Thank you for your help