Help with clustering and visualization of neurons

[zmajstorovic]

Hi, I'm reaching to you hoping that you can help me.

This is the snippet of my input data and som settings:
data= [[0.0, 0.85, 0.0, 0.0, 0.0, 0.85, 0.0, 0.85], [0.0, 0.85, 0.0, 0.9, 0.0, 0.85, 0.0, 0.9]...]
som = MiniSom(4, 4, 8, learning_rate= 0.5, neighborhood_function='gaussian', sigma=1.0)

I followed your tutorial on clustering, and instead of getting neuron-cluster representation I get neurons plotted in sets of four of the same color (total 16 neurons). I expected 16 different colors, each color for one set of the data assigned to certain neuron.
For image plotting I used the same code as in Clustering notebook

som.winner(data[0]) gets me the result e.g. (3,1)

With som.get_weights() I have following values:

Question: How do I get neuron coordinates from this table? I don't understand how (3,1) translates to coordinate system same as the input data.

[JustGlowing]

hi,

I followed your tutorial on clustering, and instead of getting neuron-cluster representation I get neurons plotted in sets of four of the same color (total 16 neurons). I expected 16 different colors, each color for one set of the data assigned to certain neuron.

Even if you have 16 neurons, during clustering only a subset might end up mapping some samples.

Question: How do I get neuron coordinates from this table? I don't understand how (3,1) translates to coordinate system same as the input data.

• You can use the method winner to get neuron coordinates.
• (3, 1) is used to define the size of the map.

My advice is to use a map with a small number of neuron first (for example (3,1)) and see if the results look like in the example, if so you can start increasing the size until it makes sense for you.