R/fuzzytocrisp.R

In mlmts: Machine Learning Algorithms for Multivariate Time Series

# Lets build a function to convert a fuzzy membership matrix to a crisp partition (maximum membership criterion)

fuzzytocrisp <- function(u){
  
  n <- nrow(u)
  c <- ncol(u)
  w <- matrix(nrow = n, ncol = c)
  
  for (i in 1 : n) {
    for (j in 1 : c) {
      
      w[i, j] = u[i, j]/max(u[i,])
    }
    
  }
  
  w[w != 1] <- 0
  clustering <- numeric(c)
  
  for (j in 1 : n) {
    
    clustering[j] <- which.max(w[j,])
    
  }
  
  clustering
}