automate scatter markers colours in groups

Hello, I have a line plot with 28 points, and I want to colour the points in groups of 2, so points 1,2, 3,4, 5,6

I started with this

            ax=app.UIAxes4;
            [x,y] = getDataFromGraph(app,ax,1);   % Get x,y, data from plot
            hold(ax,'on');
            n=numel(x);         % Typically 28, but want to allow for any (even number)
            Groups=2;           % Want to allow for 2,3,4 or 5
            
            % the number of point sis always a multiple of Groups
            
            % I started to manually write c, but its too tediuos

            c = [ 0 0 1; 0 0 1; 0 1 0; 0 1 0; 0 1 1; 0 1 1; 1 0 0; 1 0 0; 0 1 0; 1 0 0; 1 0 1; 0 1 1; 0 0 1; 0 1 0; 0 0 1; 0 1 0; 1 0 0; 1 0 1; 0 1 1; 0 0 1; 0 1 0; 0 0 1; 0 1 0; 1 0 0; 1 0 1; 0 1 1; 0 0 1; 0 1 0];
            scatter(ax,x,y, [], c, 'filled')
            

Can 'c' be automated?

(You can see the first 6 points below)

Thanks

Answer 1

[dpb]

Alternate way to define the c vector index for lookup and using a colormap...

x=1:12; y=rand(size(x));     % make up some data...
N=numel(x);                  % the number of datapoints
G=2;                         % the groups
P=N/G;                       % passes -- is assumed N evenly divisible by G; should error check
ic=kron(1:N/G,ones(1,G)).';  % create index vector into COLORS array
colormap('hot')
scatter(x,y,[],ic,'filled')

indexes into the chosen colormap by the number of groups specified...

If you want specific colors, then create them as RGB triplets for your own colormap or use index into the array row with the indexing variable as

scatter(x,y,[],COLORS(ic,:),'filled')

where COLORS is your specific color RGB array you define.

[Jason]

Perfect, thankyou!

[Jason]

Actually, is there a reason why my own map won't work:

            ax=app.UIAxes4;
            [x,y] = getDataFromGraph(app,ax,1);
            hold(ax,'on');
            N=numel(x);
            G=2;    % Groups= FPT
            passes=N/G;    % how many groups there are in the dataset to do...
            hS=gobjects(passes,1); % preallocate for scatter handles so can modify later if desired...
            % COLORS='brgcwmybrgcwmy';
            % i1=1;               % the first index to first group
            % for ix=1:passes
            %   i2=i1+Groups-1;   % the second index in group based on first and size chosen
            %   hS(ix)=scatter(ax,x(i1:i2),y(i1:i2), [], COLORS(ix), 'filled'); % and plot the points 
            %   i1=i2+1;          % update the first index for next pass
            % end
            ic=kron(1:N/G,ones(1,G)).';  % create index vector into COLORS array
            ic'
            myMap= [
              0.47 0.25 0.80; % Color 1
              0.83 0.14 0.14; % Color 2
              0.25 0.80 0.54; % Color 3
              1.00 0.54 0.00; % Color 4
              0.3960 0.5090 0.9920; % Color 5
              1 0 1; % Color 6
              0.2270 0.7840 0.1920; % Color 7
              1.0000 0.2700 0.2270; % Color 8
            ];
            colormap(ax,myMap)  %colormap(ax,'lines');  app.myColours
            scatter(ax,x,y,[],ic,'filled');
            

The number of groups is G=2, but Im getting a mixture of colour groupings

[dpb]

If I can count dots, you've got 26 points which is 13 groups but your colormap has only 8 colors. So, there aren't enough to go around...the default color maps are 256 so there are plenty for these small datasets to be unique for each group.

[Jason]

Thanks, OK I though the map would be recycled.

I gues I have to do this myself

        myMap= [
              0.47 0.25 0.80; % Color 1
              0.83 0.14 0.14; % Color 2
              0.25 0.80 0.54; % Color 3
              1.00 0.54 0.00; % Color 4
              0.3960 0.5090 0.9920; % Color 5
              1 0 1; % Color 6
              0.2270 0.7840 0.1920; % Color 7
              1.0000 0.2700 0.2270; % Color 8
              0.47 0.25 0.80; % Color 1
              0.83 0.14 0.14; % Color 2
              0.25 0.80 0.54; % Color 3
              1.00 0.54 0.00; % Color 4
              0.3960 0.5090 0.9920; % Color 5
              1 0 1; % Color 6
              0.2270 0.7840 0.1920; % Color 7
              1.0000 0.2700 0.2270; % Color 8
              0.47 0.25 0.80; % Color 1
              0.83 0.14 0.14; % Color 2
              0.25 0.80 0.54; % Color 3
              1.00 0.54 0.00; % Color 4
              0.3960 0.5090 0.9920; % Color 5
              1 0 1; % Color 6
              0.2270 0.7840 0.1920; % Color 7
              1.0000 0.2700 0.2270; % Color 8
              0.47 0.25 0.80; % Color 1
              0.83 0.14 0.14; % Color 2
              0.25 0.80 0.54; % Color 3
              1.00 0.54 0.00; % Color 4
              0.3960 0.5090 0.9920; % Color 5
              1 0 1; % Color 6
              0.2270 0.7840 0.1920; % Color 7
              1.0000 0.2700 0.2270; % Color 8
            ];

[dpb]

To do it that way, you need to remap the index lookup to the size of the map...

myMap=[
      0.47 0.25 0.80;           % Color 1
      0.83 0.14 0.14;           % Color 2
      0.25 0.80 0.54;           % Color 3
      1.00 0.54 0.00;           % Color 4
      0.3960 0.5090 0.9920;     % Color 5
      1 0 1;                    % Color 6
      0.2270 0.7840 0.1920;     % Color 7
      1.0000 0.2700 0.2270];    % Color 8
N=26; G=2;
x=1:N; y=x.^2;
ic=kron(1:N/G,ones(1,G)).';
M=height(myMap);                % will be 8 above
icc=mod(ic,M); icc(icc==0)=M;   % remap to repeat range 1:M

Then, scatter using

scatter(x,y,[],myMap(icc,:),'filled')

will produce the sets of two (G) in sequence.

BTW, the MATLAB idiom to do the above (even though not needed here) would be

myMap=repmat(myMap,8,1);       % expand to 8x original height
height(myMap)

ans = 64

[Jason]

Thanks

Answer 2

[dpb]

You need to either predefine n/Groups colors if you want them to be specific colors or you can use a linear index into a color map by group.

Then set the color as a lookup by the indexing expression based on Groups. An alternative and probably simpler coding solution would be to use hold on and multiple calls to scatter in a for...end loop; that way the color for the group is the same and just based on the index of the loop...

ax=app.UIAxes4;
[x,y] = getDataFromGraph(app,ax,1);   % Get x,y, data from plot
hold(ax,'on');
n=numel(x);         % Typically 28, but want to allow for any (even number)
Groups=2;           % Want to allow for 2,3,4 or 5
COLORS='brgckmy';   % the standard letter mnemonics can handle up to seven groups - or build custom
% the number of point sis always a multiple of Groups
passes=n/Groups;    % how many groups there are in the dataset to do...
hS=gobjects(passes,1); % preallocate for scatter handles so can modify later if desired...
i1=1;               % the first index to first group
for ix=1:passes
  i2=i1+Groups-1;   % the second index in group based on first and size chosen
  hS(ix)=scatter(ax,x(i1:i2),y(i1:i2), [], COLORS(ix), 'filled'); % and plot the points 
  i1=i2+1;          % update the first index for next pass
end

Salt to suit...

Answer 3

[Voss]

"Can 'c' be automated?"

Here is one way to automatically construct an n-by-3 matrix c with groups of repeated rows (colors), which you can use in scatter.

% somedata and axes to plot on
ax = gca();
x = 1:28;
y = 2*x.^2-5*x+3;

n = numel(x); % Typically 28, but want to allow for any (even number)
Groups = 3;   % Want to allow for 2,3,4 or 5

number_of_groups = ceil(n/Groups);

% some set of colors to use
colors = get(ax,'ColorOrder');
n_colors = size(colors,1);

% replicate the colors so there are enough for each group of points
colors = repmat(colors,ceil(number_of_groups/n_colors),1);

% construct c by appropriate indexing into the rows of colors
idx = repelem(1:number_of_groups,1,Groups);
idx = idx(1:n);
c = colors(idx,:)

c = 28×3

0 0.4470 0.7410 0 0.4470 0.7410 0 0.4470 0.7410 0.8500 0.3250 0.0980 0.8500 0.3250 0.0980 0.8500 0.3250 0.0980 0.9290 0.6940 0.1250 0.9290 0.6940 0.1250 0.9290 0.6940 0.1250 0.4940 0.1840 0.5560

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

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

% make the scatter plot
scatter(ax,x,y, [], c, 'filled')            

[Jason]

thankyou

Answer 4

[Image Analyst]

There is a function for exactly that, which makes it easy to color markers by group. It's called gscatter

help gscatter

 gscatter - Scatter plot by group
    This MATLAB function creates a scatter plot of x and y, grouped by g.

    Syntax
      gscatter(x,y,g)
      gscatter(x,y,g,clr,sym,siz)
      gscatter(x,y,g,clr,sym,siz,doleg)
      gscatter(x,y,g,clr,sym,siz,doleg,xnam,ynam)

      gscatter(ax,___)

      gscatter(___,"filled")

      h = gscatter(___)

    Input Arguments
      x - x-axis values
        numeric vector
      y - y-axis values
        numeric vector
      g - Grouping variable
        categorical vector | logical vector | numeric vector |
        character array | string array | cell array of character vectors |
        cell array
      clr - Marker colors
        MATLAB default colors (default) |
        character vector or string scalar of short color names |
        matrix of RGB triplets
      sym - Marker symbols
        '.' (default) | character vector or string scalar of symbols
      siz - Marker sizes
        positive numeric vector
      doleg - Option to include legend
        'on' (default) | 'off'
      xnam - x-axis label
        x variable name (default) | character vector | string scalar
      ynam - y-axis label
        y variable name (default) | character vector | string scalar
      "filled" - Option to fill interior of markers
        "filled"
      ax - Axes for plot
        Axes object | UIAxes object

    Output Arguments
      h - Graphics handles
        array of Line objects

    Examples
      openExample('stats/ScatterPlotWithDefaultSettingsExample')
      openExample('stats/ScatterPlotOfClimateAndHousingRatingsExample')
      openExample('stats/ScatterPlotWithMultipleGroupingVariablesAndAllOptionsExample')
      openExample('stats/SpecifyAxesForScatterPlotExample')
      openExample('stats/SpecifyMarkerColorsExample')
      openExample('stats/ModifyScatterPlotAfterCreationExample')

    See also gplotmatrix, grpstats, scatter

    Introduced in Statistics and Machine Learning Toolbox before R2006a
    Documentation for gscatter
       doc gscatter

[Jason]

OK thanks, I've taken a look - its not easy to follow for my case.

I used this:

clr = hsv(G);   % G = 2
gscatter(ax,x,y,clr)

But it needs 'idx' what ever that is.

[dpb]

 gscatter(x,y,g,clr,sym,siz,doleg,xnam,ynam)

I had forgotten there is gscatter, indeed. It has to have the grouping variable as the third argument; that could be the ic indexing vector.

Answer 5

[Jason]

Perhaps I should start a new question, apologies if I should.

The data I showed in groups of 2 has the same measurement in pairs, i.e. 1& 2 measure the same thing, 3&4 etc. I am picking up vibrations which is causing the "trianglular" spikes, so I just take the lowest value of each pair

I have just manually been removing them (the orange curve) and trying to perform smoothing or poly fit (white dashed curve)to whats left.

It would be useful to automate these removals, I have tried rmoutliers (SGolay) but it didn't touch them

Any ideas how to remove them, and ideas for more of a flat bottomed function to perform the fit (rather than the poly 4 fit)?

Howabout removing the peak in the orange curve at x=6?

Thanks

[Voss]

Here's a way to programmatically determine the lower of each pair and remove those rows. (In case of an odd initial number of rows, so that one row is unpaired, the unpaired row is preserved.)

data = readmatrix('PairedData.txt');

N = size(data,1);
if mod(N,2)
    data(end+1,:) = NaN;
    N = N+1;
end
[~,idx] = min(reshape(data(:,2),2,[]),[],1);
rows_to_keep = 2*(0:N/2-1)+idx;
data_keep = data(rows_to_keep,:);

plot(data(:,1),data(:,2),'s-')
hold on
plot(data_keep(:,1),data_keep(:,2),'x-')

[Jason]

Thankyou