How to identify outliers and remove?

Hello!

I have a .txt file with a lot of data and I need to identify and remove outliers. I already tried the command

rmoutliers

but it returns:

Undefined function or variable 'rmoutliers'.

Does anyone have any solution that can help me? I use MATLAB R2017a.

[Rik]

You will have to use a different function: that one was introduced in R2018b.

Have you tried Google? I alread see two links to file exchange submission in the sidebar (one and two). Have you looked at anything on the FEX and what functions those submissions use?

Answer 1

[Image Analyst]

pink, you can try it yourself with movmedian(), introduced in R2016a. Try this demo I made especially for you.

clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 22;
fprintf('Beginning to run %s.m ...\n', mfilename);

% Create sample noisy data.
x = linspace(0, 4*pi, 1000);
y = sin(x) + 0.1 * rand(1, length(x));
% Get 20 random indexes where we can add noise.
noiseIndexes = sort(randperm(length(x), 20));
y(noiseIndexes) = y(noiseIndexes) + 0.75 + rand(1, length(noiseIndexes));
subplot(3, 1, 1);
plot(x, y, 'b-');
grid on;
title('Original Signal', 'FontSize', fontSize);

% Smooth the signal with a median filter.
ySmooth = movmedian(y, 9);

% Get the median absolute deviation everywhere.
madValues = abs(y - ySmooth);
subplot(3, 1, 2);
plot(x, madValues, 'b-');
grid on;
title('MAD Signal', 'FontSize', fontSize);

% Find the bad indexes where MAD > 3
outlierIndexes = madValues > 0.3; % You can adjust this.
hold on;
plot(x(outlierIndexes), madValues(outlierIndexes), 'ro', 'LineWidth', 2);
grid on;

% Repair the signal by replacing the outlier locations 
% with the smoothed signal at those locations.
y(outlierIndexes) = ySmooth(outlierIndexes);
subplot(3, 1, 3);
plot(x, y, 'b-');
grid on;
title('Repaired Signal', 'FontSize', fontSize);

If that doesn't work, then attach your signal with the paper clip icon.