I need help separating some horseshoe data

Please I need help with this horseshoe data. I do not know how to separate it for a shallow neural network I have to do. Thank you.

Answer 1

[Walter Roberson]

filename = '1147406HorseShoe.csv';
data = readmatrix(filename);
Nclust = 3;
[G, centers] = kmeans(data, Nclust);

centers

centers = 3×5

13.2513 32.3622 12.5993 7.0605 19.6230 18.2635 39.1564 18.4635 8.2738 24.6921 15.9056 35.4357 14.9013 8.2562 23.1644

%let us try a visualization
scatter3(data(:,1), data(:,2), data(:,3), [], data(:,5));
hold on
scatter3(centers(:,1), centers(:,2), centers(:,3), 50, 'r', '*');
hold off

min(data)

ans = 1×5

11.2000 28.5050 9.0000 5.0000 15.7000

max(data)

ans = 1×5

20.4000 47.3650 68.2500 10.5000 28.5000

Now you can treat G as the class number for training purposes.

... for whatever good it will do you. Any semi-rational basis for chosing the class can be used since you do not have any information about what it means to "separate" the data.