test3.in

高效的k-means算法实现

  stats stage				# print summary information
  show_assignments yes			# show final cluster assignments
  validate yes				# validate assignments
  dim 2					# dimension

  data_size 10000			# number of data points
  colors 20				# ...number of clusters
  std_dev 0.01				# ...each with this std deviation
  distribution clus_gauss		# clustered gaussian distribution
  seed 1				# random number seed
gen_data_pts				# generate the data points

  kcenters 20				# number of centers
  max_tot_stage 1000 0 0 0		# number of stages
  max_run_stage 3			# number of stages per run
  min_accum_rdl 0.20			# min accumulated RDL for success

  seed 2				# use different seed
run_kmeans lloyd			# run using Lloyd's