Plotting in for loop not working. Averaging matrix over ranges.

I thought i had a code that worked but i noticed a tiny error, and ive been banging my head on a brick wall sleeplessly trying to get it to work for days.

Heres my code, i hate it, it doesn't work, if you reckon you can fix it please tell me otherwise just skip this bit and as I've decide to start again.

function TempSalplot(datafile)
datafile=importdata(datafile);
A=datafile.data;
figure;
col = flipud(parula(round(max(datafile.data(:,28)))));
%depth,distance,salinity,temp are in column 28,2,14,11 of datafile respectively;  
    for d=[1:3:max(datafile.data(:,28))]; %max is around 16
        for avd=[1:length(A(:,28))];
            if A(avd,28)>d-0.5 & A(avd,28)<d+0.5;
            B(avd,1)=A(avd,2);
            B(avd,2)=A(avd,14);
            B(avd,3)=A(avd,11);
            B(all(B==0,2),:)=[];
            end 
        [M,~,z] = unique(B(:,1),'stable');
        M(:,2) = accumarray(z,B(:,2),[],@mean);
        M(:,3) = accumarray(z,B(:,3),[],@mean);
        plot(M(:,2),M(:,3),'o-','color',col(d,:));
        hold on;  
        end
    xlabel('Salinity (psu)');
    ylabel('Tempurature (degC)');
    h=colorbar('ticks',[1:3:round(max(datafile.data(:,28)))]);
    h.Label.String=('Depth (m)');
    caxis([1,max(d)]);
    colormap(col);
end

Essentially i have 4 columns of interest say (a,b,c,d). I would like to average b and c over certain ranges of d. so I would only have as many b,c pairs for a certain d range as there are a values. Then I would like to plot that and redo it for another a range.

data=
0 15 33 0.5
0 15 32 0.8
0 16 32 1.3
0 13 34 1.6
10 13 34 0.6
10 14 35 1.0
10 16 36 1.5
10 12 33 1.9
25 13 34 0.6
25 12 33 0.9
25 12 34 1.6
25 12 36 1.9

What I think I need is a for loop.

1st iteration for 0<d=<1 (d=0.5).

M=
0 15 32.5 0.5
10 13.5 34.5 0.5
25 12.5 33.5 0.5
plot(M(:,2),M(:,3));

then repeat for d=1.5 (1<d=<2) however some of the later iteration for example (10<d=<11) have no measurements for certain values of a so should be shorter than the earlier iterations and therefore plot fewer points.

Any help on this is really appreciated , this has been a nightmare for me, thinking I had it working so many times.

[Stephen23]

I notice that you are using my solution to your earlier question:

[M,~,z] = unique(B(:,1),'stable');
M(:,2) = accumarray(z,B(:,2),[],@mean);
M(:,3) = accumarray(z,B(:,3),[],@mean);

and yet you have not accepted my answer to that question.

Answer 1

[Stephen23]

It is easy to split your data into groups according to the fourth column:

B = [...
0 15 33 0.5
0 15 32 0.8
0 16 32 1.3
0 13 34 1.6
10 13 34 0.6
10 14 35 1.0
10 16 36 1.5
10 12 33 1.9
25 13 34 0.6
25 12 33 0.9
25 12 34 1.6
25 12 36 1.9];

% Vector of the edges of each group:
D = 0:2;
% Set figure colormap befor the loop:
set(0,'DefaultAxesColorOrder',winter(numel(D)-1))
% Iterate over each group:
for k = 2:numel(D)
    % Logical index for this group:
    X = D(k-1)<B(:,4) & B(:,4)<=D(k);
    % Calculate means for this group:
    [M,~,z] = unique(B(X,1));
    M(:,4) = accumarray(z,B(X,4),[],@mean);
    M(:,3) = accumarray(z,B(X,3),[],@mean);
    M(:,2) = accumarray(z,B(X,2),[],@mean);
    % Display info (optional):
    fprintf('From %d to %d\n',D(k-1),D(k))
    disp(M)
    % Plot:
    plot(M(:,2),M(:,3))
    hold all
end

which displays this in the command window:

From 0 to 1
    0.00000   15.00000   32.50000    0.65000
   10.00000   13.50000   34.50000    0.80000
   25.00000   12.50000   33.50000    0.75000
From 1 to 2
    0.00000   14.50000   33.00000    1.45000
   10.00000   14.00000   34.50000    1.70000
   25.00000   12.00000   35.00000    1.75000

And creates this figure:

Note that I do not have the colormap parula, so I used winter instead.

[aaron Harvey]

Thank-you so much, One of the simplest and best laid out answers I've ever had! And it works perfectly. Better yet I almost understand it haha!

My code is now

function tsplot(datafilename)
A=importdata(datafilename);
B=[A.data(:,2),A.data(:,11),A.data(:,14),A.data(:,28)];
D=0:round(max(B(:,4)));
set(0,'DefaultAxesColorOrder',jet(numel(D)-1));
col=jet(numel(D)-1);
for k = 2:numel(D);                 %%%I tied changing this to k=2:5:numel(D)  
    X = D(k-1)<B(:,4) & B(:,4)<=D(k);
    [M,~,z] = unique(B(X,1));
    M(:,4) = accumarray(z,B(X,4),[],@mean);
    M(:,3) = accumarray(z,B(X,3),[],@mean);
    M(:,2) = accumarray(z,B(X,2),[],@mean);
    fprintf('From %d to %d\n',D(k-1),D(k))
    disp(M)
    plot(M(:,3),M(:,2),'o-');
    hold all
end
xlabel('Salinity (psu)');
ylabel('Tempurature (degC)');
h=colorbar;
h.Label.String=('Depth (m)');
caxis([1,max(D)]);
colormap(col);
end

however this plots every single group. How would I change it to only plot a few. I tried the changes shown above, changing the for vector. This works however the full color scheme isn't spread across the plots any more. It plots fewer lines but they are all shades of blue as opposed to the full range of jet form blue to red.

[Stephen23]

My pleasure.

I see that you are still using the code from my earlier answer, even though you have not accepted it. It is considered polite to accept answers on this forum: we volunteer our time to help you, and accepting is a small sign of appreciation from you :)

Here are some small changes that should give you want you are looking for. Firstly we can create the matrix B from A without that ugly concatenation, and you should probably use ceil instead of round:

 B = A.data(:,[2,11,14,28]);
 D = 0:ceil(max(B(:,4)));

and to create the colormap with the size that you require, try this:

V = 2:5:numel(D);
set(0,'DefaultAxesColorOrder',winter(numel(V)));
for k = V
    ... etc

Note that I got rid of col as this is not needed: you do not need to call colormap at the end of your code: the colormap has already been set by the line set(0,'DefaultAxesColorOrder'.... You can pick either of these methods, but you do not need to do both.

Avoid using the colormap jet, it is a truly awful colormap, which is explained here:

http://blogs.mathworks.com/steve/2014/10/20/a-new-colormap-for-matlab-part-2-troubles-with-rainbows/

http://www.mathworks.com/tagteam/81137_92238v00_RainbowColorMap_57312.pdf

[aaron Harvey]

Perfect! first method worked a treat. I didn't fully realise accepting and up voting weren't the same thing, I definitely 100% accept this answer!! Its not just fixed the code but taught me what the fix was too. You have helped a panicing student claw back a huge chunk of his final project.

Answer 2

[Eng. Fredius Magige]

Hi You might review few lines, specially for and if, as to be:

for d=1:3:max(datafile.data(:,28)) %max is around 16 for avd=1:length(A(:,28)) if (d<A(avd,28)+0.5 && d>A(avd,28)-0.5)

try it

[aaron Harvey]

unfortunately no that didn't work. The vairable B is not being found outside the if function? but I'm not sure my nested for and ifs work anyway.