Separate and Operate on Cells Based on Label

Hello,

I need assistance with separating out a cell array consisting of text labels within each cell based on the label. I have tried the findgroups function, but I would like to have the labels remain for further operations instead of the generated numbers for the labels (for example, I'd want to work with group "bicep" instead of group "1"). If I am misunderstanding how findgroups works and that is a possible outcome with that function, please explain how, as I am still new to Matlab.

Secondly, once those cells are separated based on the label, I need some way to go through all of the groups separately with a function (in this case, a mean of the unique values in corresponding cells). I understand how to do the function if all values were in the same group, but I am confused if working with numerous groups at the same time.

If there's anything else needed to help with my case, please ask and I can give more information. Thank you. (see comments)

[Azzi Abdelmalek]

Your question is not precise.

[Brian Tomblin]

1. Given varying and repeating text labels in the cells of a 1xN cell array, how do you group the cells based on the labels? Preferably by still keeping the labels instead of changing them to something else (like a number reference).

2. How do you apply a function (like unique) to separated groups at the same time so as not to have to do it individually for each group?

Is that better?

[per isakson]

Could you provide a tiny example of input and required output.

Have you thought of using Create and Work with Tables? (Introduced in R2013b)

[Brian Tomblin]

Ex. Run program with file; gives 2 outputs: a Legend for the Objects and the Objects

Legend = [Biceps, Triceps, Forearm, Chest, Deltoid]
ObjectLabels = [Biceps, Deltoids, Biceps, Chest, Chest, Forearm, Forearm, Triceps]
ObjectXposition = [1,3,2,3,1,4,2,3]
ObjectYposition = [1,3,2,3,1,4,2,3]
ObjectZposition = [1,3,2,3,1,4,2,3]

So I need to plot the Objects separately based on their label while keeping the cell so that the corresponding values can also be used. Something like:

    Biceps Locations   Deltoid Loc.    Chest Loc.
    [Biceps, Biceps]    [Deltoid]    [Chest, Chest]
          [1,2]           [3]             [3,1]
          [1,2]           [3]             [3,1]
          [1,2]           [3]             [3,1]

I already know how to plot such a setup, so how can I separate/sort the list of labels? I believe that connecting it to the Legend would be easiest, but I am unsure how to do that.

[per isakson]

"I'd want to work with group "bicep" instead of group "1"" &nbsp means that you you want to use statements like

plot( ..., 'bicep' )

How large is the total dataset, typically?

[Brian Tomblin]

It can vary depending on how much data is given. Currently working with around 14 labels. A way to group the data such that naming each individual label during plotting would be unnecessary would be nice, if possible. Along the idea that the groups are separated, but still remain under a variable LABELS or something like that, so plot(Labels) would go through each group individually and plot them without having to name them. If that's not a possibility, it can be worked around.

Answer 1

[Image Analyst]

Try the ismember() function. To get the means, you can use grpstats().

[Brian Tomblin]

So this is the idea that I am looking for (checking one list for corresponding values from another list), but it seems like the ismember() only gives a true/false for if any repetitions at all are there instead of giving a means to separate the corresponding values. Am I mistaken in the ismember() capabilities or is there another way to do that?

[Stephen23]

Have a look at ismember's second output. You could use that output in accumarray to group/process your data together.

[Brian Tomblin]

Using the Lia or Lia,Locb function requires a string instead of cells, which doesn't work due to the long list of text that ends up in a single cell. I cannot seem to find the members when all of them are together in the same cell. Possibly text was not intended for the ismember() function.