mloc_som

 MLOC_SOM Create an mloc object and train its SOM
  
   CALLS
       mloc = mloc_som(y,nq,p,np)
       mloc = mloc_som(y,nq,p,np,dims,epochs,Nc)
  
   INPUTS
         y: Output data matrix (Ry x Q)
        nq: delays of y in the index vector (endogenous part of the index vector)
         p: exogenous variables of the index vector (Rp x Q)
        np: delays of p in the index vector (exogenous part of the index vector)
      dims: size of the SOM (by default {30,30})
    epochs: number of epochs (by default, 10)
        Nc: final neighborhood (by default, 0.4)

      Note: a typical index vector has the following structure
               phip = [y(k-n1) y(k-n2)... y(k-nq), p(k-n1)... p(k-np)]
                       ------------------------    ------------------
                            endogenous part           exogenous part


   OUTPUTS
      mloc: Structure of the SOM with local models defined on the space generated by phip

   DESCRIPTION
      Create an mloc object and train its SOM in the "embedded" space. 
      The embedded space contains a set of variables (selectors of dynamics) 
      that are supposed to define the different dynamic behaviours of
      the system. 
 
      There is an "endogenous part" (previous system outputs, 
      meaning that the system dynamic behavior depend on its own state), 
      and an "exogenous part" (known external variables, meaning that 
      the system dynamic behavior depends on external factors)

   SEE ALSO
      MLOC_LOC, MLOC_SIM, MLOC_PLANOS, MLOC_COMPARA