How to group variable of interests by logical indexing?

I have this sort of data. I want to plot PAW with Pasture 9 Patch 1, Pasture 9 with Patch 3, along with Pasture 9 with Patch 5. I want to use logical indexing as, Pasture (Pasture==9), Patch(Patch==1) and then retrieve PAW data based on it. I tried this,

for ii =1:length(PAW)
    Pasture_index = find(Pasture==9);
    PAW40pasture(ii) = mean(PAW40(Pasture_index,1));
end

But this did not work. Any suggestions/help will be appreciated.

Pasture  Patch  PAW
9  1  59.90366347
9  1  58.71485861
9  1  57.35723603
9  1  57.36212973
9  3  57.04769458
9  3  127.7284367
9  3  143.7897127
9  3  143.5349223
9  3  142.5156813
9  5  142.3629269
9  5  148.7025447
9  5  149.7302998
9  5  147.5149873
9  5  88.55115557
9  5  80.52936618

[Jan]

I've formatted the code using the "{} Code" button.

Please explain "did not work" with any details. It is easier to solve a problem than to guess what teh problem is. I do not understand the description: "plot PAW with Pasture 9 Patch 1, Pasture 9 with Patch 3, along with Pasture 9 with Patch 5".

[Sonisa]

It works but the problem is instead of getting 3 values, I am still getting 15 values. I should get PAW data by 1, 3, and 5 patches but I am not getting it. Hope it will help you to understand Jan Simon.

[Image Analyst]

What is the data type of Pasture? Is it a double, a cell array, or a table? How many rows and columns does it have? Same question for PAW.

[Sonisa]

all are double. Three column and 16 rows.

[Sonisa]

Image Analyst, I sent you my file. Please have a look on it

Answer 1

[Guillaume]

First, you're not using logical indexing in your code because of the find. The exact same code without the call to find (i.e. Pasture_index = Pasture==9;) would be using logical indexing.

It's not clear to me why you thought the snippet you wrote would produce the result you want since you never filter for a given patch value.

If all you want is to produce the mean of PAW for identical combinations of pasture and patch then you don't need a loop and just need accumarray:

[ppval, ~, rows] = unique([Pasture, Patch], 'rows'); %assuming that Pasture and Patch are column vectors
PAW40pasture = accumarray(rows, PAW40(:, 1), [], @mean);
out = array2table([ppval, PAW40pasture], 'VariableNames', {'Pasture', 'Patch', 'mean'}) %this line just for pretty display

If you wanted to plot them (against what x axis?) then you could do it with a loop as follows:

figure; hold on;
ppvalues = unique([Pasture, Patch]); 
for ppvalue = ppvalues'
   inpasturepatch = Pasture == ppvalue(1) & Patch == ppvalue(2); %This is logical indexing. No find
   plot(PAW40(inpasturepatch, 1), 'DisplayName', sprintf('Pasture %d, Patch %d', ppvalue(1), ppvalue(2)));
end

[Sonisa]

Thanks a lot

Answer 2

[Image Analyst]

Do you have the stats toolbox? It looks like your data is in a table. You might be able to get the mean for all numbers very easily in one line of code with grpstats:

statarray = grpstats(tbl,groupvar)

[Sonisa]

It doesn't work. Do you mind if I sent you data of how it looks like?

[Image Analyst]

Sure, you can attach it to your original question with the paper clip icon.

[Guillaume]

In base matlab, there is now more or less the same with the findgroups and splitapply workflow.

Or you can use accumarray

[Sonisa]

plot command didnot work, it says Index exceeds matrix dimensions.

[Image Analyst]

You accepted an answer and said thanks, so I'm assuming you got everything working and this doesn't apply anymore.

[Sonisa]

I have attached data and I want to group data by Pasture 9, patch 1 and date, Pasture 9, Patch 3 and date, and Pasture 9, Patch 5 and date. I used the grpstats to create groupvar. But it is giving me data as the original file which is 1795*4 dimension. Actually i want it to be 598*4 dimension. Hope it makes sense to you.

Thanks so much.