Neural Clustering Tool Testing Output

Hello, I have been using the neural clustering tool in Matlab 2020b. I have trained my neural network with 80% of my data, and then tested the network with the remaining 20% of my data. I can see the SOM Sample Hit and other plots from my test, but how do I obtain an output matrix from the test to understand where each observation was placed within the Sample Hit plot?

Answer 1

[Yash Sharma]

Hi Sydney,

To obtain and analyze the output matrix for your test data in a Self-Organizing Map (SOM) in MATLAB 2020b, follow these steps:

1. Get Network Outputs for Test Data:

outputs = net(testData);

2. Convert Outputs to Winning Neuron Indices: winningNeurons contains indices of the winning neurons for each test observation, indicating their placement on the SOM grid .

winningNeurons = vec2ind(outputs);

3. Visualize the SOM Hit Map:

plotsomhits(net, testData);

This visualizes where each observation is placed on the SOM grid.

These steps provide both the indices of winning neurons for analysis and a visual representation of the test data distribution on the SOM.

Hope it helps!