Automatically select and name variables from import file (.xlsx)

Hello,

I have an excel file with a column of variables. It looks like this:

The sequence is approx. 0.0237, -0.0016, 0.0018, 0.0239. Which should be named A, B, C, D. This sequence repeats in the same order 50 times (A1,B1,C1,D1, A2, B2, C2, D2, etc), but the values are different each time and the number of cells for A1, B1, etc varies. A and D are always in the range of 0.0001 to 0.0320, while B and C are always in the range 0.0010 to 0.0020.

At the moment, I am highlighting the cells from Import Data and naming them one by one.

I would like to automatically select 10 cells before B1, and name those A1. Automatically select all the cells for B1 and name them B1. Automatically select all the cells for C1 and name them C1. Finally, automatically select 10 cells after C1 and name them D1. This would be repeated for the other 49 times.

Any help would be appreciated!

Thank you.

[Mathieu NOE]

hello

1/ could you share the excel file ?

2/ what is the range of B ? there must be a typo in your text, as you mention D twice and B never :

A and D are always in the range of 0.0001 to 0.0320, while C and D are always in the range 0.0010 to 0.0020.

[meryem berrada]

Yes, thank you for noticing my typo! I corrected it and added the excel file.

I appreciate your help!

[Mathieu NOE]

Hello again

so I tried to understand your logic - how you split the data into the four groups

if I'm right , these blocks correspond to the four "plateau" phases that are here repeated several times.

But right now , I doubt that using only the criteria of min / max values can be enough and robust as one can see for exxample , the "B" section can be either positive and negative, and also the min / max values for each A/B/C/D group are not "excluding" - I mean the valid ranges do overlap.

I am more thinking about slecting portion of data that have very low slope (definitioon of a "plateau") with other constrains like min number of consecutive samples and distance to the previous section

[J. Alex Lee]

can you take second derivative to find the "corners" and split on them?

[meryem berrada]

Thank you for your answers. If it helps, we can make B always the negative sequence (so B and C would be interchangeable).

Would there be a way to:

• find the first deviation from the mean, name it B1
• then, select the 10 values before B1, name it A1
• select all the values between B1 and the next change, name it C1
• select 10 values after C1 and name it D1

Or, is there a way to do highlighting the cells manually in the Import data folder and press enter (for example), and the selected values would automatically be named A1 ? Then I can highlight the next set of values and press enter and it would automatically be named B1, etc.

Thank you,

Answer 1

[Mathieu NOE]

ok so that's my first trial , based on slope values (must be below a given threshold to select sections of interest)

not sure I got all 10 points as per request , but so far i am not sure how to improve my code

results :

code :

clc
clearvars
data = importdata('Example file MatLab.xlsx');
data = data.data;

xx = 1:length(data);
dy = gradient(data,1);
ind = find(abs(dy)<0.5e-5); % results may vary according to slope threshold

figure(1);
subplot(211),plot(xx,data,'b',xx(ind),data(ind),'r+');
subplot(212),plot(dy); 


ck = 1;
k = 1;
out = [];
out(1) = ind(1);
for ci = 2:length(ind)
    ind_delta = ind(ci)-ind(ci-1);
    if (ind_delta-1)<eps
        k = k+1;
        out(k,1) = ind(ci);
    else
%          if k>=1
            out2{ck} = out;
            out = [];
            k = 1;
            ck = ck+1;
%          end
    end

end
    % last buffer 
    out2{ck} = out;
    
    
% expected result is 40 arrays (out2)
nb_arrays = numel(out2);
indA  = (out2((1:4:nb_arrays)))';
indB  = (out2((2:4:nb_arrays)))';
indC  = (out2((3:4:nb_arrays)))';
indD  = (out2((4:4:nb_arrays)))';

indAA=cell2mat(indA);
indAA(indAA==0) = [];

indBB=cell2mat(indB);
indBB(indBB==0) = [];

indCC=cell2mat(indC);
indCC(indCC==0) = [];

indDD=cell2mat(indD);
indDD(indDD==0) = [];

figure(2);
plot(xx,data,'b',xx(indAA),data(indAA),'r+',...
    xx(indBB),data(indBB),'g+',...
    xx(indCC),data(indCC),'m+',...
    xx(indDD),data(indDD),'c+');
legend('data','A sections','B sections','C sections','D sections');

[meryem berrada]

Thank you so much that's really helpful! The outputs indAA, indBB, wtc are all the locations of the sequences in x-axis. Is there a way to get the corresponding values in y-axis ?

[Mathieu NOE]

My pleasure

see that the x values are given for A segments by xx(indAA) and the corresponding y values are data(indAA)

same applies for B / C / D with indBB , indCC and indDD

[meryem berrada]

Perfect, thank you so much!

[Mathieu NOE]

you're welcome !