Avoiding zeros doesnt wrok while plotiing from a data file.

x=load(sprintf('test_line.dat'))
x1=x(:,1);
x1r=x(:,2);
 % x1(x1r==0)=Nan %This shows erorr
x1r(x1r==0)=NaN;
x10=plot(x1,x1r,--b','LineWidth', 1.5);
%x20=plot(x1,x1r,.g','LineWidth', 1.5);
hold on

As long its line plot, it doenst work and connect next point with a line. If I use x1(x1r==0)=NAN, it still shows line connecting a gap. Using 'dot'as '.g' in the coommnet works fine, but if I want to plot many such plots overlapping, that will show only one color. Any sugegstions please?. Line plots would be betetr, if i can plot. Thanks...

Edit: Thes size of x1 or x1r =1450088 and that of nonzeros are x1(x1r~=0)=133604, if that is creating any problem (connecting gap while plotting nonzeros).

[Steven Lord]

Are you certain that the elements of x1r are exactly zero or are they only close to zero?

x = [1e6 2.1 3e6 4.2]

x = 1×4

1.0e+06 * 1.0000 0.0000 3.0000 0.0000

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

While the second and fourth elements of x are small enough relative to the first and third elements that they display as 0.0000 they are not in fact equal to 0 as you can see if you extract them.

[x(2), x(4)]

ans = 1×2

2.1000 4.2000

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

If you wanted to make anything "small enough" equal to NaN, you can. In this case, I'm defining 10 as "small enough".

x(abs(x) < 10) = NaN

x = 1×4

1000000 NaN 3000000 NaN

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[Anil]

Dear @Steven Lord, thank for that. I understand your point. If i plot with a mark '.',, it shows a gap between two points that makes sense that there are no points in between; generally connected by a line.

Answer 1

[Mathieu NOE]

hello

one method is to use interpolation to fill the voids (NaN)

% Example data with NaN values
x = 1:10;
y = sin([1, 2, NaN, 4, 5, NaN, NaN, 8, 9, 10]/2);

% Interpolate to fill NaN values
y_filled = interp1(x(~isnan(y)), y(~isnan(y)), x, 'linear', 'extrap');

% plot the result
plot(x,y,'*',x,y_filled)

[Mathieu NOE]

Alternative Methods :

1/ Using fillmissing:

% Example data with NaN values
x = 1:10;
y = sin([1, 2, NaN, 4, 5, NaN, NaN, 8, 9, 10]/2);
% Interpolate to fill NaN values
y_filled = fillmissing(y, 'linear'); %This is a simpler and more direct way to fill NaN values using linear interpolation.
% plot the result
plot(x,y,'*',x,y_filled)

2/ Using inpaint_nans : If you need more advanced handling of NaN values, you can use the inpaint_nans function available on MATLAB File Exchange : inpaint_nans - File Exchange - MATLAB Central

[Anil]

Dear @Mathieu NOE, Thanks for that. But I nedd not to fill but skip such points where zeros appear. Marks such as '*','.' and 'other' works fine for a single plot. But if i have 4 or 5 overlapping plots and these marks don't make any snense. Thats why I am curious about line plots with different style (to show overlaps) but avoiding zeros. Thanks.

[Mathieu NOE]

hello Anil

it would probably help me if you could share your data and the code and also a sketch of what you try to achieve

if your data has NaN (as you did) the line plot will just not show this portion of the plot. Isn't it what you wanted in first place ?

% Example data with NaN values
x = 1:10;
y = sin([1, 2, NaN, 4, 5, NaN, NaN, 8, 9, 10]/2);
% plot the result
plot(x,y)

●

[Mathieu NOE]

maybe this could be of some interest for you : here I plot several sine waves with some NaN cretad in the y range -0.1 / + 0.1

just in case you are looking for ideas about colrs and line styles (there are tons of available info's and ressources about going beyong standard matlap plotting styles)

your are not obliged to use markers like I did below

t = 0:0.05:2*pi;
MarkerStyle={'o','+','*','x','s','d','v','>'};

% create some new highly distinguishable colors
Colors=[1 0 0;
    1 0 1;
    0 1 0;
    0.4660 0.6740 0.1880;
    0 0 1;
    0.3010 0.7450 0.9330;
    0.8500 0.3250 0.0980;
    0.9290 0.6940 0.1250];

figure(1),hold on 
for ck = 1:8
    y = sin((1+0.2*ck)*t);
    y(abs(y)<0.1) = NaN;
    p = plot(t,y,'LineWidth',2,'Color',Colors(ck,:),'Marker',MarkerStyle{ck});
    leg_str{ck} = (['Line #' num2str(ck)]);
end
legend(leg_str);

●

[Anil]

Dear @Mathieu NOE Thanks for that. I am sharing a data file, where I want to plot x(:,1) against x(:,17)........x(:,21) (total 5 line plots). Ooops, zip file size goes above 5MB. How can I share it?

[Mathieu NOE]

try with WeTransfer or equivalent (share the link)

[Anil]

Dear @Mathieu NOE, Thank for that. please find a link here

[Mathieu NOE]

hello again

try this code below

NB that the last five columns of the x array that you want to plot are numerically equal so the 5 lines will overlay perfectly (if this is waht you wanted ?)

this is a screenshot of my text editor showing the last columns of your file :

this is what I got , once the rows of y data filled with zeroes are removed :

also I used a trick to reduce the amount of data plotted as your file has a ton of rows (maybe the sampling rate is quite high for the job). for a good plot around 1000 points can suffice , no need to have zillions of points.

clc
clearvars

x=readmatrix('test_line.dat');
x1=x(:,1);
y=x(:,17:21);

% now select data for which no row of y is filled with zeros 
rowCheck = all(y >eps, 2);
x1 = x1(rowCheck,:);
y = y(rowCheck,:);

% lest downsample a bit the data
% take every r samples to plot 
samples = size(x1);

r = 100; % downsampling factor
ind = (1:r:samples);
x1 = x1(ind,:);
y = y(ind,:);

figure
plot(x1,y);

[Anil]

Dear @Mathieu NOE, Thanks for that. i have that plot. what i want is line plots but with a gap near -13.5 to -11.5, a flat line connecting two end points should be absent. You can see this (there are no points in between) using any mark '.' insetad of using a line for the plots.

[Mathieu NOE]

ok, now I see what you want ! :)

see the last figure

this is what I get on my side :

now if you want the same color for the same curve (left and right portions ) this is also doable..

clc
clearvars

x=readmatrix('test_line.dat');
x1=x(:,1);
y=x(:,17:21);

% now select data for which no row of y is filled with zeros 
rowCheck = all(y >eps, 2);
x1 = x1(rowCheck,:);
y = y(rowCheck,:);

% lest downsample a bit the data
% take every r samples to plot 
samples = size(x1);

r = 100; % downsampling factor
ind = (1:r:samples);
x1 = x1(ind,:);
y = y(ind,:);

figure
plot(x1,y);

% now split the data and plot together the 2 parts
[val,indm] = max(diff(x1)); % find the x discontinuity

figure
plot(x1(1:indm),y(1:indm,:));% left block 
hold on 
plot(x1(indm+1:end),y(indm+1:end,:));% right block 
hold off

[Mathieu NOE]

a small improvement in the code to show the plots having same color on both sides

to actually show the 5 differents curves (that are normally overlaid) , I added a vertical offset in the form of +0.01*k in those lines :

    plot(x1(1:indm),y(1:indm,k)+0.01*k,'Color',colors(k,:));% left block 
    plot(x1(indm+1:end),y(indm+1:end,:)+0.01*k,'Color',colors(k,:));% right block 

plot looks like now :

full code :

clc
clearvars

x=readmatrix('test_line.dat');
x1=x(:,1);
y=x(:,17:21);

% now select data for which no row of y is filled with zeros 
rowCheck = all(y >eps, 2);
x1 = x1(rowCheck,:);
y = y(rowCheck,:);

% lest downsample a bit the data
% take every r samples to plot 
samples = size(x1);

r = 100; % downsampling factor
ind = (1:r:samples);
x1 = x1(ind,:);
y = y(ind,:);

figure
plot(x1,y);

% now split the data and plot together the 2 parts
[val,indm] = max(diff(x1)); % find the x discontinuity

% define my prefered colors 
%% define your custom color order

% option 1 : manually : possible but takes time and not flexible / robust
colors = [0 0 1;...
          0 1 0;...
          1 0 0;...
          0 1 1;...
          1 0 1;...
          1 0.69 0.39;...
          0.6 0.2 0;...
          0 0.75 0.75;...
          0.22 0.44 0.34;...
          0.32 0.19 0.19]; %10x3 RGB array

% % or based on existing color maps 
% colors = jet(n);

% or this 
% % https://fr.mathworks.com/matlabcentral/fileexchange/42673-beautiful-and-distinguishable-line-colors-colormap
% colors = linspecer(60);

% % even better : 
% func = @(x) colorspace('RGB->Lab',x); % https://fr.mathworks.com/matlabcentral/fileexchange/28790-colorspace-transformations
% colors = distinguishable_colors(n,{'w','k'},func); % https://fr.mathworks.com/matlabcentral/fileexchange/29702-generate-maximally-perceptually-distinct-colors


figure
hold on
for k = 1:size(y,2)
    plot(x1(1:indm),y(1:indm,k)+0.01*k,'Color',colors(k,:));% left block 
    plot(x1(indm+1:end),y(indm+1:end,:)+0.01*k,'Color',colors(k,:));% right block 
end
hold off

[Anil]

Dear @Mathieu NOE, thank you very very much for your efforts and help. I will try this one. Thanks

[Anil]

Dear @Mathieu NOE, I slighlty modified to show the overlapping lines as follow:

linS = {'-','--',':','-.',':'};

l2={'A','B','C','D','E'};

lw={4.3,4.0,3.7,3.4,3.2};

for k = 1:size(y,2)

x0(k)=plot(x1(1:indm),y(1:indm,k),'Color',colors(k,:),'linestyle',linS{k},'LineWidth', lw{k});% left block hold on;

hold on

plot(x1(indm+1:end),y(indm+1:end,:),'Color',colors(k,:),'linestyle',linS{k},'LineWidth', lw{k});% right block

hold on

end

legend(x0([1, 2, 3, 4, 5]),{'$A$','$B$','$c$','$D$','$E$'},'Fontsize',46,'Location','Northwest', 'interpreter', 'latex');

legend('boxoff')

Thank you very much.

[Mathieu NOE]

ok good

tx for accepting my answer

have a good day

fyi you are not obliged to repeat "hold on" inside the for loop , simply before the for loop is enough

this should give you the same result :

figure
hold on
linS = {'-','--',':','-.',':'};
 l2={'A','B','C','D','E'};
 lw={4.3,4.0,3.7,3.4,3.2};
for k = 1:size(y,2)
      x0(k) = plot(x1(1:indm),y(1:indm,k),'Color',colors(k,:),'linestyle',linS{k},'LineWidth', lw{k});% left block  
      plot(x1(indm+1:end),y(indm+1:end,:),'Color',colors(k,:),'linestyle',linS{k},'LineWidth', lw{k});% right block 
end
legend(x0([1, 2, 3, 4, 5]),{'$A$','$B$','$c$','$D$','$E$'},'Fontsize',46,'Location','Northwest', 'interpreter', 'latex');
legend('boxoff')
hold off