Controlling colors in scatter plots

I have an array of label names that have corresponding RGB triplet colors:

% label names
myLabel = ["tomatoes";"carrots";"blueberries";"beans";"bananas"];
myLabelColor = [0.9 0.0 0.0; 1.0 0.7 0.1; 0.4 0.0 0.8; 0.3 0.7 0.0; 0.9 0.9 0.1];

I want to use these custom colors such that the first element of myLabel is always colored with the first element of myLabelColor in my scatter plots and graphs.

Here is some data:

x = rand(10,1);
y = rand(10,1);
groups = ["bananas"; "blueberries"; "bananas"; "beans"; "beans"; "tomatoes";"blueberries";"bananas";"tomatoes";"tomatoes"];
figure;
gscatter(x,y, groups);

Note that the "carrots" label is missing from this data but may appear in another set of data at some point. I don't know which labels will appear in each data set but I know they will always be ones from myLabel array

In the scatter plot, the groups are identified by Matlab's default colors. How can I edit my code so that the groups are identified by my custom colors (bearing in mind that I will want to scale up to larger data sets that may or may not contain all labels)?

Answer 1

[DGM]

This is what I threw together, but there's probably a better way to do this; I haven't really used gscatter() for anything before.

% fake data
x = rand(10,1);
y = rand(10,1);
groups = ["bananas"; "blueberries"; "bananas"; "beans"; "beans"; "tomatoes";"blueberries";"bananas";"tomatoes";"tomatoes"];

% label naming info
myLabel = ["tomatoes";"carrots";"blueberries";"beans";"bananas"];
myLabelColor = [0.9 0.0 0.0; 1.0 0.7 0.1; 0.4 0.0 0.8; 0.3 0.7 0.0; 0.9 0.9 0.1];

% gscatter() will list groups based on order of appearance
% figure out which colors are used and the order they're listed
ugroups = unique(groups,'stable');

% then create a new color table ordered to fit the groups as they appear
[islabeled,idx] = ismember(ugroups,myLabel);
if any(~islabeled) % just to avoid confusion
    error('grouping variable contains groups which have no assigned label');
end
CT = myLabelColor(idx,:);

% plot it
hs = gscatter(x,y,groups,CT);

[SM1974]

Thanks very much, @DGM

[SM1974]

Is there a way to increase the marker size in the legend without changing the marker size in the plot?

[SM1974]

Don't worry. I found the marker size answer here: https://uk.mathworks.com/matlabcentral/answers/348946-how-to-set-legend-marker-size

[DGM]

For future readers, the behavior of legend() has been changing and a lot of old answers may be problematic. Walter's comment is the tip of the iceberg. See also Adam's answer here:

https://www.mathworks.com/matlabcentral/answers/50250-how-can-i-change-the-marker-size-in-legend

Star Strider and I did a bit of a summary review here:

https://www.mathworks.com/matlabcentral/answers/2160035-what-can-i-do-to-make-the-pattern-in-the-legend-larger-i-look-up-a-lot-of-ways-can-t-solve-ask-for

The short story is that many of these answers rely on an undocumented (iirc deprecated) output from legend() which may cause unexpected formatting problems if its requested. The form of the descendant objects that make up the legend items has also changed over the years, so that changes how one would search for them and/or customize them.