How to merge and smooth multiple curves in a plot, if there are discrete points?

I have a graph, containing 4 different types of curves, in 5 frequency regions, overall 20 curves in a single graph. At some points, the curves meet, however at some points, the curves are super distant from the following one. I applied brushing, but still have discontinuities.

(Keeping the color coding) I need to merge them and smooth the sharp and points. How can I do this?

I uploaded the fig file. I have the excel data if needed.

Best

( 1. I failed to utilize link in figure properties 2. I tried to gather the data in excel, and delete the meeting points, it also didnt work)

[Star Strider]

Having the Excel file, tthe code you used to plot those, and some idea of what the original data represent (what they are and how they were created) would help.

[Image Analyst]

Inserting screenshots directly into your post with the image icon is most helpful. After that, attaching PNG files with the paperclip icon is next best. Attaching Fig files is least convenient because we now need to switch out of our browser into MATLAB to read them in and display them.

If you have any more questions, then attach your data, and code to read it in and make the plots, with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

There may be a nan in your data and nan's will cause that point not to be plotted.

[Paul]

No need to switch from browser to Matlab to read and display fig files. Can insert files from this thread into a comment or answer (Paperclip -> Link from this thread) and then

openfig('unbrushed')

ans = 
  Figure (1) with properties:

      Number: 1
        Name: ''
       Color: [1 1 1]
    Position: [1 49 1536 740.8000]
       Units: 'pixels'

  Use GET to show all properties

though would be better if OP did this.

[Zeynep Ertekin]

Happy to get an answer, however i dont understand what you mean :(

Can we talk on the brushed version if possible?

[Image Analyst]

@Paul thanks for letting us know that - I didn't know. However it's still extra work - we need to write and run code - when it would be easier on us if the original poster just made it displayed right from the start.

@Zeynep Ertekin Not sure what you don't understand or what poster you were referring to. Did you not understand how to attach your data file and m-file with the paperclip icon when I requested that you do that? Or was it something else you didn't understand? Regardless, we're still waiting for your data so we can show you how to do it. It might be as simple as locating NaNs and removing them, like I already said.

badIndexes = isnan(x) | isnan(y);
x(badIndexes) = []; % Remove the bad x values.
y(badIndexes) = []; % Remove the bad y values.

[Zeynep Ertekin]

Dear Image Analyst,

I add here my workspace. I appreciate your help.

Answer 1

[Mathieu NOE]

hello

tried afew options, probably here my best result in the very limited time I have now :

this is a demo on the first two line plots , extend to the other plots if you like it

load('dataread_matlab.mat')
% figure
% plot(TEM_34_S{:,3},-1*TEM_34_S{:,4},'r',TEM_62_S{:,3},-1*TEM_62_S{:,4},'r',TEM_90_S{:,3},-1*TEM_90_S{:,4},'r',TEM_159_S{:,3},-1*TEM_159_S{:,4},'r',TEM_284_S{:,3},-1*TEM_284_S{:,4},'r','LineWidth',4);
% hold on
% plot(TEM_34_S{:,7},-1*TEM_34_S{:,8},'b',TEM_62_S{:,7},-1*TEM_62_S{:,8},'b',TEM_90_S{:,7},-1*TEM_90_S{:,8},'b',TEM_159_S{:,7},-1*TEM_159_S{:,8},'b',TEM_284_S{:,7},-1*TEM_284_S{:,8},'b','LineWidth',4);
% hold on

%%
channel = 4;
x1 = [TEM_34_S{:,channel-1};TEM_62_S{:,channel-1};TEM_90_S{:,channel-1};TEM_159_S{:,channel-1};TEM_284_S{:,channel-1}];
y1 = [TEM_34_S{:,channel};TEM_62_S{:,channel};TEM_90_S{:,channel};TEM_159_S{:,channel};TEM_284_S{:,channel}];

%%
channel = 8;
x2 = [TEM_34_S{:,channel-1};TEM_62_S{:,channel-1};TEM_90_S{:,channel-1};TEM_159_S{:,channel-1};TEM_284_S{:,channel-1}];
y2 = [TEM_34_S{:,channel};TEM_62_S{:,channel};TEM_90_S{:,channel};TEM_159_S{:,channel};TEM_284_S{:,channel}];

%% main loop
[x1s,y1s] = mysmoother(x1,y1);
[x2s,y2s] = mysmoother(x2,y2);

figure
plot(x1,y1,'*r',x1s,y1s,'g',x2,y2,'*b',x2s,y2s,'y','LineWidth',4);

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [xc,ys] = mysmoother(x,y)

% take unique values 
[x,ia,~] = unique(x);
y = y(ia);

% my smoother
N = 100;
xi = linspace(min(x),max(x),N); %  

% create median average in each bin
for k=1:N-1
    ind = (x>=xi(k) & x<xi(k+1));
    xc(k) = (xi(k) + xi(k+1))/2;
    yc(k) = median(y(ind));
end
% add first and last points (assumed to be in a "clean" area)
xc = [x(1) xc x(end)];
yc = [y(1) yc y(end)];

% smooth it 
ys = smoothdata(yc,'rloess',round(N/3));

end

[Zeynep Ertekin]

Dear Mathieu NOE ,

I appreciate your perfect answer. Even though the code worked, I failed to get what you mean by channel. Can you please explain just a little bit.

Thank you in advance

Best of the best,

[Mathieu NOE]

hello @Zeynep Ertekin

I needed to store each curve as x, y arrays simply and avoid calling / repeating a long sequence each time like

TEM_34_S{:,col-1};TEM_62_S{:,col-1};TEM_90_S{:,col-1};TEM_159_S{:,col-1};TEM_284_S{:,col-1} for one x vector.

I simply used your "pattern" and changing the column index to have x and y

if you'd wanted this now could be turned into a for loop for even shorter and better code (in the case for example you have to process many files / or more variables)

here some new code for the 4 plots now :

load('dataread_matlab.mat')

%% replace original code 
% figure
% plot(TEM_34_S{:,3},-1*TEM_34_S{:,4},'r',TEM_62_S{:,3},-1*TEM_62_S{:,4},'r',TEM_90_S{:,3},-1*TEM_90_S{:,4},'r',TEM_159_S{:,3},-1*TEM_159_S{:,4},'r',TEM_284_S{:,3},-1*TEM_284_S{:,4},'r','LineWidth',4);
% hold on
% plot(TEM_34_S{:,7},-1*TEM_34_S{:,8},'b',TEM_62_S{:,7},-1*TEM_62_S{:,8},'b',TEM_90_S{:,7},-1*TEM_90_S{:,8},'b',TEM_159_S{:,7},-1*TEM_159_S{:,8},'b',TEM_284_S{:,7},-1*TEM_284_S{:,8},'b','LineWidth',4);
% hold on

% plot(wg_34_S{:,3},-1*wg_34_S{:,4},'c',wg_62_S{:,3},-1*wg_62_S{:,4},'c',wg_90_S{:,3},-1*wg_90_S{:,4},'c',wg_159_S{:,3},-1*wg_159_S{:,4},'c',wg_284_S{:,3},-1*wg_284_S{:,4},'c','LineWidth',4);
% hold on
% plot(wg_34_S{:,7},-1*wg_34_S{:,8},'g',wg_62_S{:,7},-1*wg_62_S{:,8},'g',wg_90_S{:,7},-1*wg_90_S{:,8},'g',wg_159_S{:,7},-1*wg_159_S{:,8},'g',wg_284_S{:,7},-1*wg_284_S{:,8},'g','LineWidth',4);


%% with this one
figure
col_indexes = [4,8];

for k = 1:numel(col_indexes)
    col = col_indexes(k);
    % first datas
    x1 = [TEM_34_S{:,col-1};TEM_62_S{:,col-1};TEM_90_S{:,col-1};TEM_159_S{:,col-1};TEM_284_S{:,col-1}];
    y1 = [TEM_34_S{:,col};TEM_62_S{:,col};TEM_90_S{:,col};TEM_159_S{:,col};TEM_284_S{:,col}];
    % second datas
    x2 = [wg_34_S{:,col-1};wg_62_S{:,col-1};wg_90_S{:,col-1};wg_159_S{:,col-1};wg_284_S{:,col-1}];
    y2 = [wg_34_S{:,col};wg_62_S{:,col};wg_90_S{:,col};wg_159_S{:,col};wg_284_S{:,col}];
    % pass x1,y1 & x2,y2 to mysmoother function
    [x1s,y1s] = mysmoother(x1,y1);
    [x2s,y2s] = mysmoother(x2,y2);
    plot(x1,y1,'*',x1s,y1s,'-',x2,y2,'*',x2s,y2s,'-','LineWidth',4);
    hold on

end

hold off

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [xc,ys] = mysmoother(x,y)
% take unique values 
[x,ia,~] = unique(x);
y = y(ia);
% my smoother
N = 100;
xi = linspace(min(x),max(x),N); %  
% create median average in each bin
for k=1:N-1
    ind = (x>=xi(k) & x<xi(k+1));
    xc(k) = (xi(k) + xi(k+1))/2;
    yc(k) = median(y(ind));
end
% add first and last points (assumed to be in a "clean" area)
xc = [x(1) xc x(end)];
yc = [y(1) yc y(end)];
% smooth it 
ys = smoothdata(yc,'rloess',round(N/3));
end

[Zeynep Ertekin]

I do not know, how to thank enough. I see that you named the columns as channel to simplfy the stuff.

You are the best.

Wish you success and goodness all the time (fingers crossed)

[Mathieu NOE]

Tx for the nice comments , but I still have to work hard if I want to access the top10 league !!!

have a great day !