how can ? convert to execute python k-means algorythm to x-means and k-means algorthm ?

SelimSelim Member Posts: 32 Contributor I
edited June 18 in Help
hello everybody , ı have doing k-means clustering via execute python operator but ı want to apply to this process k-medoid and x-means so ı need to change to k-means code section in my python screen so what ı need to write code for x-means and k-medoid to k-means execute python screen .ı have shared my python screen 
-------------code----------------
import pandas as pd
from operator import itemgetter
import numpy as np
import random
import sys
from scipy.spatial import distance
from sklearn.cluster import KMeans


C = 10
def k_means(X) : 
  kmeans = KMeans(n_clusters=C, random_state=0).fit(X)
  return kmeans.cluster_centers_

def samevolumecluster( D, volumes):
    """ in: point-to-cluster-centre distances D, Npt x C
            
        out: xtoc, X -> C, equal-size clusters
       
    """
    
    clustervolume = (np.sum(volumes)) / C

    xcd = list( np.ndenumerate(D) )  # ((0,0), d00), ((0,1), d01) ...
  
    xcd.sort( key=itemgetter(1) )

    xtoc = np.ones( Npt, int ) * -1

    nincluster = np.zeros( C, int )

    vincluster = np.zeros( C, int)

    nall = 0

    for (x,c), d in xcd:
        
        if xtoc[x] < 0 and vincluster[c] <= clustervolume+10:
 
            xtoc[x] = c

            nincluster[c] += 1
            
            vincluster[c] += volumes[x]

            nall += 1

            if nall >= Npt:  break

    return xtoc

def rm_main(data):
 
  data_2 = data.values
  volumes = data_2[:,1]

  centres = k_means(data_2)
  D = distance.cdist( data_2, centres)
  xtoc = samevolumecluster( D, volumes)
  data['cluster'] = xtoc
  return data
Tagged:

Answers

  • David_ADavid_A Moderator, Employee, RMResearcher, Member Posts: 177  RM Research
    Hi,

    did you know that there is no need to code these clustering algorithms by yourself.
    RapidMiner has a lot of clustering methods as build in, ready to use, operators.

    Best,
    David


    SGolbert
Sign In or Register to comment.