I want to only interpolate shorter amounts of data and leave the longer amounts of NaN not interpolated.
2 visualizzazioni (ultimi 30 giorni)
Mostra commenti meno recenti
I have interpolated my data by using:
nanp = isnan(data);
t = 1:numel(data);
data(nanp) = interp1(t(~nanp), data(~nanp), t(nanp), 'linear');
The issue is all of the data gets interpolated, including big chunks of data that I would rather leave empty. Is there a way to make guidelines for what data gets interpolated? Such as "If more than x NaN's in a row, then do not interpolate that chunk of data".
0 Commenti
Risposte (4)
kaan ilter
il 26 Gen 2022
As I understand you would like to remove some outliers. How about removing some values by using the following filter?
datanew=data(data>9999,:)=[];
0 Commenti
Matt J
il 26 Gen 2022
Modificato: Matt J
il 26 Gen 2022
If more than x NaN's in a row, then do not interpolate that chunk of data
Using function from here,
this can be done with,
G=groupTrue(nanp);
[~,~,runlengths]=groupLims(G,1);
doInterp=ismember( G,find(runlengths<=x));
data(dointerp) = interp1(t(~nanp), data(~nanp), t(doInterp), 'linear');
0 Commenti
Walter Roberson
il 26 Gen 2022
max_length_to_interpolate = 17; %change as appropriate
nanp = isnan(data(:).'); %must have row vector
starts = strfind([false nanp], [false true]);
stops = strfind([nanp false], [true false]);
mask = (stops - starts + 1 <= max_length_to_interpolate);
starts = starts(mask); stops = stops(mask);
if ismember(1, starts) || ismember(numel(data), stops)
error('asked to interpolate leading or trailing nan, which we cannot do');
end
nan_replacements = arrayfun(@(B,E) interp1(t([B-1,E+1]), data([B-1,E+1]), t(B:E), 'linear'), starts, stops, 'uniform', 0);
for K = 1 : length(starts)
data(starts(K):stops(K)) = nan_replacements{K};
end
if really needed it would be possible to do the final assignments without a loop.
0 Commenti
Image Analyst
il 27 Gen 2022
This works. It uses the bwareaopen() function in the Image Processing Toolbox to throw out stretches of nans longer than you specify.
% Create sample data.
y = [10,20,30,nan, nan, 60,70,80, nan, nan, nan, nan, 90,100,110]
% Get the indexes we're supposed to interpolate over.
xOriginal = 1 : length(y);
% Find nan locations
nanMask = isnan(y)
% Define the longest stretch of nans you'll interpolate over.
nanLength = 3;
% Get mask of stretches over that length.
longNanMask = bwareaopen(nanMask, nanLength) % Needs Image Processing Toolbox.
% Get non-nan data:
x2 = xOriginal(~nanMask);
y2 = y(~nanMask)
% Interpolate over everything, both long and short nan runs.
v2 = interp1(x2, y2, xOriginal)
% Replace long nan stretches with nans
v2(longNanMask) = nan
fprintf('Done!\n');
y =
10 20 30 NaN NaN 60 70 80 NaN NaN NaN NaN 90 100 110
nanMask =
1×15 logical array
0 0 0 1 1 0 0 0 1 1 1 1 0 0 0
longNanMask =
1×15 logical array
0 0 0 0 0 0 0 0 1 1 1 1 0 0 0
y2 =
10 20 30 60 70 80 90 100 110
v2 =
10 20 30 40 50 60 70 80 82 84 86 88 90 100 110
v2 =
10 20 30 40 50 60 70 80 NaN NaN NaN NaN 90 100 110
Done!
0 Commenti
Vedere anche
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!