What is a good way (other than ~=) to represent that a variable does not contain NaN values in R2013?

My current code is aiming to perform basic arithmetic on a variable, but only if it contains values. I don't want to necessarily remove null values, but just find a good way of saying if my variable, data, actually contains some threshold of values, it can continue to do this basic arithmetic. My current code is as follows:

   if data ~= isnan
                average_length = 150;
                means = [];
                datalength =  size(data,1);
                speed =data(:,1);
                insta_rate = data(:,2);
                for count = 1:average_length:datalength
                    if count+average_length > datalength
                        mean_speed = mean(speed(count:datalength));
                        mean_insta_rate = mean(mean(count:datalength));
            end
      end 
end

Thanks! :)

Answer 1

[Geoff Hayes]

Katherine - what do you mean by contains some threshold of values? If all values are NaN then obviously there is nothing to do, but do you wish to do "something" if at least one row in your data matrix does not include NaNs in both columns?

Note that your if condition won't work because there is no input argument to the isnan function, and comparing a matrix (data) with another in a if condition is not really what you want either.

I know that you said that you don't want to necessarily remove NaN values, but without them there you would be able to some work once the threshold has been applied

 % define a threshold (so if at least 75% of rows are not-NaN, then use data)
 threshold = 0.75;

 % determine those rows of data that don't have a NaN in both columns
 nonNanRows =  ~isnan(data(:,1)) & ~isnan(data(:,2));

 if sum(nonNanRows)/length(data) > threshold

    % reset data to that which has the non-NaN data
    data = data(nonNanRows,:);

    % continue with your code to extract the speed, etc.
 end

Try the above and see what happens!