How can I set every value in a row to NaN based on two non-consecutive Quality columns?

Hello, I am trying to throw out suspicious data from my .csv file based on two quality indicators. The file is (112,000 x 5) with the first column as datetime data, the second and fourth wind direction and wind speed respectively and the third and fifth their respective quality indicators (1 = good, 0 = bad).

I would like to set every value in a row (except the datetime) to NaN, if any one of the quality indicators is zero. How can that be done in the most efficient manner?

I tried to follow this community post (How do I delete an entire row if a specific column contains a zero? - MATLAB Answers - MATLAB Central (mathworks.com). I read the data in as wind with the function 'dataset', and then neither 'logical' nor the operator '==' worked for some reason.

wind = dataset('file','C:\my\directory\Example_data.csv', ...
     'delimiter',',','format',['%s' repmat(' %f',1,4)]); 
wind.Properties.VarNames = [{'Date'} {'Direction'} {'Quality1'} {'Velocity'} {'Quality2'}]; %new headers

wind((:,[3 5])==0) = NaN;

I work with Matlab R2018b.

[Jiri Hajek]

Hi, note that datetime class oes not have NaN, there is a NaT instead. But there's also a glitch in the assignment

wind((:,[3 5])==0) = NaN;

which probably does not contain correct logical indices. Seems like you rather wanted to write this:

wind(wind(:,[3 5])==0) = NaN;

[Nils]

Yes, but I had also already tried this and I get the Error: "Undefined operator '==' for input arguments of type 'dataset'."

Answer 1

[Sajid Afaque]

first try to get the indices where qny of the quality indicator is 0 (i.e. find the row numbers where either column 3 or column 5 is 0)

then you can set columns of that row to NaN

PLEASE TRY THE CODE BELOW

quality1_index = find(wind.Kvalitet1 == 0);
quality2_index = find(wind.Kvalitet2 == 0);
final_indices = unique([quality1_index;quality2_index]);
wind.Kvalitet1(final_indices) = NaN;
wind.Kvalitet2(final_indices) = NaN;

line 1 and line 2 can be further optimized and written as below

final_indices = find(wind.Kvalitet1 == 0 |wind.Kvalitet2 == 0 ); %quality1_index & quality2_index are found together

[Nils]

Thank you, that is almost exactly what I want. How do I go from here to setting columns 2 and 4 to NaN as well, if either column 3 or 5 are NaN?

[Sajid Afaque]

nan_indices = find(isnan(wind.Kvalitet1) | isnan(wind.Kvalitet2)); %find in which rows column 3 or 5 has NaN
wind.Vindriktning(nan_indices) = NaN;
wind.Vindhastighet(nan_indices) = NaN;

Please dont forget to accept the answer, as it would help others in similar situtation as you

[Nils]

Works perfectly, you made my day!