group numbers in cells

Hi,

I have a 1*5 cell and each of them contains three columns and a variable numer of rows.

In this case, the first column of each cell represents my temperture and the other two columns contain data. As mentioned, each cell has a different number of rows, i.e., the temperatures are not lying in the same row number.

Here what I would like to do: find the same temperature in each cell and group them in a unique array of data, along to the rest of the data, i.e. the content of the other two columns.

Thanks for your help

[Yongjian Feng]

Try to write something first, and we can help you to improve.

1. Show us sample data
2. Show how you handle it.
3. Show the error if it failed

[Alessandro]

please find attached part of the code.

By loading the two .m files, I have 1*2 cell.

What I would like is a new series of 1*N cells that contains:

1) 11 45.88 0.7510

2) 12 65.91 0.92

12 43.57 0.16

3) 13 ... etc

4) 15 ...

The 'grouping' variables are the numbers in the first column.

I have been trying different ways but I have no clue how to handle it.

Could you please give advice on this?

files = dir('data*.mat');
table = struct2table(files);
sample_numfiles = length(files);
sample_data_rh = cell(1, sample_numfiles);

for k = 1:sample_numfiles
    
    sample_data_rh{k} = load(files(k).name);
    rh_data{k} = (sample_data_rh{k}.RH(:,:));
    
end

Answer 1

[Dave B]

I think you're asking how to go from cells that have a sort of group-id in the first column to recombined cells that all have that value in the first column?

Here's one idea (this felt a little convoluted but I think it does the trick):

• combine everything into one matrix
• sort it
• use findgroups to convert those group identies into consecutive integers
• use histcounts to get counts (i.e. how many values in each new cell)
• use mat2cell to convert it back to a cell

d1=load('data1');
d2=load('data2');
c = {d1.RH d2.RH};
alldata=cell2mat(c(:));
alldata=sortrows(alldata,1);
grpid=findgroups(alldata(:,1));
res=mat2cell(alldata, histcounts(grpid,'BinWidth',1), 3)';
res{1}

ans = 1×3

11.0000 45.8899 0.7510

res{2}

ans = 2×3

12.0000 65.9147 0.9272 12.0000 43.5758 0.1654

res{3}

ans = 2×3

13.0000 62.5894 0.3408 13.0000 40.1022 0.0644

[Alessandro]

Thanks Dave B, that works.

However, I have an additional question. Is there any way to differentiate where the results in res{1}, res{2}... res{n} come from the different .mat files? I am asking this beceause, in reality, I need to do this job with more than 2 .mat files (each .m file corresponds to my 'sample concentration').

Roughly, a result like this would be ideal:

data 1 13.0000 62.5894 0.3408

data 2 13.0000 40.1022 0.0644

data n 13.0000 20.3001 0.0267

thanks!

[Dave B]

I think you want a table, because this looks like a table (a bunch of rows with heterogenous data). I'd question whether you really want to split it up into separate cells, perhaps it's better to keep everything in one table and just generally plan on being able to index out a particular temperature (that's seems like a more MATLABby approach IMO).

d1=load('data1');
d2=load('data2');
c = {d1.RH d2.RH};

t=table;
for i = 1:numel(c)
    % Note: if you only have your cell array for the purposes of loading
    % files, just load your files here (i.e. loop over file names rather
    % than putting them in a cell in the first place!)
    this_t=array2table(c{i},'VariableNames',["Temperature" "Thing1" "Thing2"]);
    this_t.Source(1:height(this_t))="data" + i;
    t=[t;this_t];
end

t=sortrows(t,["Temperature" "Source"]);
t=movevars(t,'Source',"Before",'Temperature'); % order doesn't really matter but this looks prettier
head(t)

ans = 8×4 table
    Source     Temperature    Thing1     Thing2 
    _______    ___________    ______    ________

    "data2"         11         45.89     0.75101
    "data1"         12        65.915     0.92723
    "data2"         12        43.576     0.16542
    "data1"         13        62.589     0.34078
    "data2"         13        40.102    0.064415
    "data1"         15        56.666     0.26838
    "data2"         15        36.772     0.49241
    "data1"       17.5        50.386     0.62024

% Suggest you stop here, you can always iterte over temperatures or data
% sources etc. So why bother with splitting it up? It will make nothing
% easier and some things harder...

% if you really want to split it up:
grps = unique(t.Temperature);
separates = cell(1,numel(grps));
for i = 1:numel(grps)
    separates{i} = t(t.Temperature==grps(i),:);
end
separates{1}

ans = 1×4 table
    Source     Temperature    Thing1    Thing2 
    _______    ___________    ______    _______

    "data2"        11         45.89     0.75101

separates{2}

ans = 2×4 table
    Source     Temperature    Thing1    Thing2 
    _______    ___________    ______    _______

    "data1"        12         65.915    0.92723
    "data2"        12         43.576    0.16542

separates{3}

ans = 2×4 table
    Source     Temperature    Thing1     Thing2 
    _______    ___________    ______    ________

    "data1"        13         62.589     0.34078
    "data2"        13         40.102    0.064415

[Alessandro]

Super!!

Thank you so much for your help, Dave.