1

trainSOM() takes proto0 parameter as input. Can prototype from already trained SOM be passed while training new data?

Trying to reuse the code vector/prototype. I am trying below code :


    Brero1.som <- trainSOM(x.data = DATA, nb.save=2,
                          dimension = c(5,5),
                           radius.type="letremy",
                           dist.type="letremy",
                             maxit=maxit_var,
                           eps0=0.0100000,init.proto="random", scaling ="unitvar")
    
    Brero2.som <- trainSOM(x.data = New_DATA, nb.save=2,
                           dimension = c(5,5),
                           radius.type="letremy",
                           dist.type="letremy",
                             maxit=maxit_var,proto0=Brero1.som$prototypes
                           eps0=0.0100000,init.proto="random", scaling ="unitvar")

Error: Error in run.trainSOM(x.data, ...) : initial prototypes dimensions do not match SOM parameters: in the current SOM, prototypes must have <XX> rows and <XX> columns

I have tried different sizes for the prototypes. Has anyone tried this proto0 ??

Sneha
  • 11
  • 1

0 Answers0