How to cluster data with multiple classes with SOM and colour the map

My data has 1700 rows (samples) and 186 predictors. I have 52 different groups or classes (Grp 1, 2, 5, 23, etc up to 52) to cluster with Self Organising Maps(SOM). I am using the syntax selforgmap in Matlab:

MyMap= selforgmap(dimensions,coverSteps,initNeighbor,topologyFcn,distanceFcn)

I am not sure how to input the dimension example 10x10, coversteps 200 etc in the syntax above

I also need help with a code to capture all the different classes in the SOM map with various colours if this is possible

Thanks

Answer 1

[Amey Waghmare]

Hi,

As per my understanding, you want to cluster a data using ‘Self organizing map’ of dimension 10x10 and cover steps of 200. You can use the following syntax to create the SOM with these properties as follows,

MyMap =  selforgmap([10 10], 200, 3, 'hextop', 'linkdist')

This will create a self organizing map with the following properties,

• dimensions of the SOM: [10 10]
• coverSteps: 200
• initNeighbor: 3
• topologyFcn: ‘hextop’
• distanceFcn: ‘linkdist’

For more information on ‘selforgmap’, please refer to the following link, https://in.mathworks.com/help/deeplearning/ref/selforgmap.html

You can also use ‘Neural Net Clustering’ app to cluster the data using Self organizing maps. Please refer to the following example which explains the creation of self organizing maps and their visualizations,

https://in.mathworks.com/help/deeplearning/gs/cluster-data-with-a-self-organizing-map.html

You can plot using the functions such as plotsomhits, plotsomnc, plotsomnd, plotsomplanes, plotsompos, and plotsomtop. For more information on these functions, please refer to the documentation link below,

https://in.mathworks.com/help/deeplearning/self-organizing-maps.html

I hope this helps.