How to plot with t-SNE in MATLAB from 5 elements composition into 2 dimensions?

Question

Majid Vaghari 2022-2-17

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1652625-how-to-plot-with-t-sne-in-matlab-from-5-elements-composition-into-2-dimensions

回答： Akanksha 2025-3-1

t-SNE plot.JPG

I would like to make a figure like the below figure!

T-distributed stochastic neighbor embedding (t-SNE) used in the figure to visualize across the composition space in two dimensions (I have a 5 element composition space for the alloy and entropy of mixing calculate from the composition).

Entropy of mixing=;

We have the data for the entropy of mixing and compositions in the Excel attached file.

My question is now how to visualize from composition space with 5 elements into just two dimensions! The results should be like a pentagon, because the attached picture is for a 6 elements alloy and is like a hexagon.

P.mat attached matrix contains x1,x2,x3,x4,x5,.

How to make a t-SNE from P matrix?

Best

Majid

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Majid Vaghari 2022-2-17

The data is big so I cannot attached them here.

If anyone can solve the problem, I can send the data by email.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Akanksha 2025-3-1

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1652625-how-to-plot-with-t-sne-in-matlab-from-5-elements-composition-into-2-dimensions#answer_1560916

在 MATLAB Online 中打开

Hi Majid Vaghari,

The resolution to visualize from composition space with 5 elements into 2D can be achieved using the following code :

%1. Load your Nx5 data matrix: P. Also, ensure P is in your MATLAB workspace. 
%2. Run t-SNE  
Y = tsne(P, 'Perplexity', 30, 'Exaggeration', 5); 
%3. Plot the results, coloring by S (an Nx1 vector) 
figure; 
scatter(Y(:,1), Y(:,2), 50, S, 'filled'); 
colorbar; 
xlabel('t-SNE 1'); 
ylabel('t-SNE 2'); 
title('t-SNE Projection of 5-element Compositions'); 

I have tested this code on small sample data set and successfully generated a meaningful 2D scatter plot, where each point is coloured by its associated property value.