How to split data into several segments so that in each segment, the number of NaN is not more than a certain value.

I have a large data which contains around 6,500,000+ data points. I need to split/divide my data into several segments so that in each segment, the number of consecutive NaN is not more than 18,000. What should I try? Thanks.

[Rik]

Is that the only requirement? If so, you can simply use isnan and a bit of code.

Answer 1

[Image Analyst]

Try this (just off the top of my head):

data = rand(1, 6500000);
segmentLength = 1000000; % Whatever...
nansPerSegment = 30000; % Whatever...
for k = 1 : segmentLength : length(data)
  lastIndex = min(length(data), k + segmentLength - 1);
  thisData = data(k:lastIndex);
  numElements = numel(thisData);
  nanLocations = k + randperm(numElements, nansPerSegment) - 1;
  data(nanLocations) = nan;
end