Deleting 'NaN' rows out of a table

Hi everybody!

I got: A table (18316x4). Within the last two columns I got strings representing time (saved in cells). So sometimes instead of a date, here I can find a string called "NaN".

I'd like to: Remove those "NaN" rows I can find in my 3rd column, so later on I can transform my string-dates into real datetimes.

What I got so far: Only the idea of using the function "isnan" (though here I get the error message, that it is undefined for input arguments of type cell OR that I got too many input arguments)or the function "find" ... Which doesn't work either.

nan1=t(all(isnan(t.r(:,1)),3),:);  

nan2=find(t.r{:,1});

My question: How can I easily remove the "NaN"-Strings out of my table, or better say, the whole row where I can find one?

I appreciate all help I get! Best regards.

Answer 1

[Walter Roberson]

Convert to datetime first. The nan will become NaT. Then use https://www.mathworks.com/help/matlab/ref/rmmissing.html

[jrbbrt]

Thank you for your quick answer Walter!

If I try converting my 3rd column to datetime, this works perfectly. But if I try to do so with the 4th column ...

date=datetime(t.r(:,1),'Inputformat','dd.MM.yyyy HH:mm:ss');

Unfortunately I'm getting an error message: "Error using datetime (line 639) Input data must be a numeric array, a string array, a cell array containing character vectors, or a char matrix."

And I rlly don't know why. My very first guess was, that those NaN's might be the reason ...

What do you think?

[Paolo]

If the data contains milliseconds like for your previous question, you forgot to include the sss specified in datetime.

Use:

date = datetime(t.r(:,1),'InputFormat','dd.MM.yyyy HH:mm:ss.SSS')

[Peter Perkins]

It's likely that t.r(:,1) isn't what you expect it to be. What is it? Perhaps you can show t.r(1:5,1).