How do I remove NaN and / or interpolated data?

Hello.

I have the vectors

A=[1 NaN 3  ;...
   4 NaN 6  ;...
   5 NaN NaN;...
   0 NaN NaN;...
   7 8   NaN;...
   0 0   9] ;...

(variable A) and t=0:1:5; (time), and I would like to do the following: If in a column there are more than 4 rows with NaN I exclude this column, otherwise I interpolate this data.

I have no idea how to do this. Can anyone help me giving suggestions?

Thanks.

[madhan ravi]

Give an example of your output.

Answer 1

[Guillaume]

A=[1 NaN 3  
   4 NaN 6  
   5 NaN NaN
   0 NaN NaN
   7 8   NaN
   0 0   9] 

%remove columns with more than 4 nans
A(:, sum(isnan(A), 1) >= 4) = [];

%replace nans by interpolated values. Requires R2016b or later
A = fillmissing(A, 'linear')

[Thaís Fernandes]

The 'fillmissing' function did not run on my version of my matlab. So I used the first part that you suggested and then the suggestion of another person on another question (see below) to interpolate.

Thank you for your help. :)

 

%To interpolate, I used
nanx = isnan (A);
  t = 1: numeral (A);
  A (nanx) = interp1 (t (nanx), A (nanx), t (nanx));