Basic_Exaples_20160908.ipynb

Awesome work!! Thank you so much.

Hi,
Is there any way so that i can have same cluster number on same dataset every time i run, like a seed parameter?

extract data rows that match desired clusters after clusterization

Is it possible to extract coordinates of all features in the map? Let say, we have 100 features to generate a SOM map with 10x10 clusters, how to know the coordinates of features shown in the map? thank you so much.

Absolutely, you can enter into the source program and acquire the data what you want though add a few lines of code at the end of som.train() function, Actually, all of the graphics must be drawn by computed data,you can let that data return to you in source code. hope to help

I am trying to install sompy but getting the error 'cannot import name 'SOM' from 'sompy' (/anaconda3/lib/python3.7/site-packages/sompy/init.py'

Please help

Is there any prediction example?
By passing input determining the cluster..?

I am trying to install sompy but getting the error 'cannot import name 'SOM' from 'sompy' (/anaconda3/lib/python3.7/site-packages/sompy/init.py'

Please help

Any solution? I'm facing the same issue.

Did you install sompy?

Create new environment in python and download the code from repo.
go to your python environment and traverse to downloaded code and run the script
python setup.py install

I tried that and it didn't work. I finally installed it following the next steps:

1. pip install SOMPY-master.zip
2. pip install ipdb

Is there any documentation somewhere? It's difficult to use some methods without documentation or examples.

Example:
https://gist.github.com/sevamoo/035c56e7428318dd3065013625f12a11

I think we have only this example..
We need to read the code..
There some more SOM libraries, in case you need them
minisom : https://github.com/JustGlowing/minisom/blob/master/examples/Iris.ipynb (well documented, author available on twitter)
neupy : Self-Organizing Feature Maps (SOFM or SOM)
somoclu and SimpSOM

Hello,
I would like to create a map not only with the trained instances, but also with unseen data that are going to be used by the function project_by(I assume). Is there a way to do something like this? Thanks

Indeed, good question